diff --git a/csv/_io.ts b/csv/_io.ts
index 4fefbcfa9e2d..346818ab72a3 100644
--- a/csv/_io.ts
+++ b/csv/_io.ts
@@ -45,7 +45,8 @@ export interface ReadOptions {
    * If negative, no check is made and records may have a variable number of
    * fields.
    *
-   * If the wrong number of fields is in a row, a `ParseError` is thrown.
+   * If the wrong number of fields is in a row, a {@linkcode SyntaxError} is
+   * thrown.
    */
   fieldsPerRecord?: number;
 }
@@ -227,11 +228,13 @@ export function createQuoteErrorMessage(
 export function convertRowToObject(
   row: string[],
   headers: readonly string[],
-  index: number,
+  zeroBasedLine: number,
 ) {
   if (row.length !== headers.length) {
     throw new Error(
-      `Error number of fields line: ${index}\nNumber of fields found: ${headers.length}\nExpected number of fields: ${row.length}`,
+      `record on line ${
+        zeroBasedLine + 1
+      } has ${row.length} fields, but the header has ${headers.length} fields`,
     );
   }
   const out: Record<string, unknown> = {};
diff --git a/csv/mod.ts b/csv/mod.ts
index d24467dee631..5afc62b948fc 100644
--- a/csv/mod.ts
+++ b/csv/mod.ts
@@ -56,7 +56,9 @@
  *
  * results in
  *
+ * ```ts no-assert
  * [`the "word" is true`, `a "quoted-field"`]
+ * ```
  *
  * Newlines and commas may be included in a quoted-field
  *
diff --git a/csv/parse.ts b/csv/parse.ts
index 4c4b0e15eb85..bc2bed875261 100644
--- a/csv/parse.ts
+++ b/csv/parse.ts
@@ -208,7 +208,7 @@ class Parser {
     this.#input = input.startsWith(BYTE_ORDER_MARK) ? input.slice(1) : input;
     this.#cursor = 0;
     const result: string[][] = [];
-    let _nbFields: number | undefined;
+
     let lineResult: string[];
     let first = true;
     let lineIndex = 0;
@@ -225,6 +225,25 @@ class Parser {
       throw new Error("Invalid Delimiter");
     }
 
+    // The number of fields per record that is either inferred from the first
+    // row (when options.fieldsPerRecord = 0), or set by the caller (when
+    // options.fieldsPerRecord > 0).
+    //
+    // Each possible variant means the following:
+    // "ANY": Variable number of fields is allowed.
+    // "UNINITIALIZED": The first row has not been read yet. Once it's read, the
+    //                  number of fields will be set.
+    // <number>: The number of fields per record that every record must follow.
+    let _nbFields: "ANY" | "UNINITIALIZED" | number;
+    if (options.fieldsPerRecord === undefined || options.fieldsPerRecord < 0) {
+      _nbFields = "ANY";
+    } else if (options.fieldsPerRecord === 0) {
+      _nbFields = "UNINITIALIZED";
+    } else {
+      // TODO: Should we check if it's a valid integer?
+      _nbFields = options.fieldsPerRecord;
+    }
+
     while (true) {
       const r = this.#parseRecord(lineIndex);
       if (r === null) break;
@@ -234,19 +253,15 @@ class Parser {
       // the number of fields in the first record
       if (first) {
         first = false;
-        if (options.fieldsPerRecord !== undefined) {
-          if (options.fieldsPerRecord === 0) {
-            _nbFields = lineResult.length;
-          } else {
-            _nbFields = options.fieldsPerRecord;
-          }
+        if (_nbFields === "UNINITIALIZED") {
+          _nbFields = lineResult.length;
         }
       }
 
       if (lineResult.length > 0) {
-        if (_nbFields && _nbFields !== lineResult.length) {
+        if (typeof _nbFields === "number" && _nbFields !== lineResult.length) {
           throw new SyntaxError(
-            `record on line ${lineIndex}: wrong number of fields`,
+            `record on line ${lineIndex}: expected ${_nbFields} fields but got ${lineResult.length}`,
           );
         }
         result.push(lineResult);
@@ -269,7 +284,7 @@ export interface ParseOptions {
    * are ignored. With leading whitespace the comment character becomes part of
    * the field, even you provide `trimLeadingSpace: true`.
    *
-   * @default {"#"}
+   * By default, no character is considered to be a start of a comment.
    */
   comment?: string;
   /** Flag to trim the leading space of the value.
@@ -319,11 +334,25 @@ export interface ParseOptions {
  * @example Usage
  * ```ts
  * import { parse } from "@std/csv/parse";
- * import { assertEquals } from "@std/assert";
+ * import { assertEquals } from "@std/assert/equals";
  *
- * const string = "a,b,c\nd,e,f";
+ * const string = "a,b,c\n#d,e,f";
  *
- * assertEquals(parse(string), [["a", "b", "c"], ["d", "e", "f"]]);
+ * assertEquals(parse(string), [["a", "b", "c"], ["#d", "e", "f"]]);
+ * ```
+ *
+ * @example Quoted fields
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const string = `"a ""word""","comma,","newline\n"\nfoo,bar,baz`;
+ * const result = parse(string);
+ *
+ * assertEquals(result, [
+ *   ['a "word"', "comma,", "newline\n"],
+ *   ["foo", "bar", "baz"]
+ * ]);
  * ```
  *
  * @param input The input to parse.
@@ -333,26 +362,139 @@ export function parse(input: string): string[][];
 /**
  * Parses CSV string into an array of objects or an array of arrays of strings.
  *
- * If `column` or `skipFirstRow` option is provided, it returns an array of
+ * If `columns` or `skipFirstRow` option is provided, it returns an array of
  * objects, otherwise it returns an array of arrays of string.
  *
- * @example Usage
+ * @example Don't skip first row with `skipFirstRow: false`
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const string = "a,b,c\nd,e,f";
+ * const result = parse(string, { skipFirstRow: false });
+ *
+ * assertEquals(result, [["a", "b", "c"], ["d", "e", "f"]]);
+ * assertType<IsExact<typeof result, string[][]>>(true);
+ * ```
+ *
+ * @example Skip first row with `skipFirstRow: true`
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const string = "a,b,c\nd,e,f";
+ * const result = parse(string, { skipFirstRow: true });
+ *
+ * assertEquals(result, [{ a: "d", b: "e", c: "f" }]);
+ * assertType<IsExact<typeof result, Record<string, string | undefined>[]>>(true);
+ * ```
+ *
+ * @example Specify columns with `columns` option
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const string = "a,b,c\nd,e,f";
+ * const result = parse(string, { columns: ["x", "y", "z"] });
+ *
+ * assertEquals(result, [{ x: "a", y: "b", z: "c" }, { x: "d", y: "e", z: "f" }]);
+ * assertType<IsExact<typeof result, Record<"x" | "y" | "z", string>[]>>(true);
+ * ```
+ *
+ * @example Specify columns with `columns` option and skip first row with
+ * `skipFirstRow: true`
  * ```ts
  * import { parse } from "@std/csv/parse";
- * import { assertEquals } from "@std/assert";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
  *
  * const string = "a,b,c\nd,e,f";
+ * const result = parse(string, { columns: ["x", "y", "z"], skipFirstRow: true });
+ *
+ * assertEquals(result, [{ x: "d", y: "e", z: "f" }]);
+ * assertType<IsExact<typeof result, Record<"x" | "y" | "z", string>[]>>(true);
+ * ```
+ *
+ * @example TSV (tab-separated values) with `separator: "\t"`
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const string = "a\tb\tc\nd\te\tf";
+ * const result = parse(string, { separator: "\t" });
+ *
+ * assertEquals(result, [["a", "b", "c"], ["d", "e", "f"]]);
+ * ```
+ *
+ * @example Trim leading space with `trimLeadingSpace: true`
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const string = " a,  b,    c\n";
+ * const result = parse(string, { trimLeadingSpace: true });
+ *
+ * assertEquals(result, [["a", "b", "c"]]);
+ * ```
+ *
+ * @example Lazy quotes with `lazyQuotes: true`
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const string = `a "word","1"2",a","b`;
+ * const result = parse(string, { lazyQuotes: true });
+ *
+ * assertEquals(result, [['a "word"', '1"2', 'a"', 'b']]);
+ * ```
+ *
+ * @example Set comment prefix with `comment` option
+ * ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const string = "a,b,c\n# THIS IS A COMMENT LINE\nd,e,f";
+ * const result = parse(string, { comment: "#" });
+ *
+ * assertEquals(result, [["a", "b", "c"], ["d", "e", "f"]]);
+ * ```
+ *
+ * @example Infer the number of fields from the first row with `fieldsPerRecord: 0`
+ *  ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertThrows } from "@std/assert/throws";
+ *
+ * // Note that the second row has more fields than the first row
+ * const string = "a,b\nc,d,e";
+ * assertThrows(
+ *   () => parse(string, { fieldsPerRecord: 0 }),
+ *   SyntaxError,
+ *   "record on line 2: expected 2 fields but got 3",
+ * );
+ * ```
+ *
+ * @example Enforce the number of fields for each row with `fieldsPerRecord: 2`
+ *  ```ts
+ * import { parse } from "@std/csv/parse";
+ * import { assertThrows } from "@std/assert/throws";
  *
- * assertEquals(parse(string, { skipFirstRow: false }), [["a", "b", "c"], ["d", "e", "f"]]);
- * assertEquals(parse(string, { skipFirstRow: true }), [{ a: "d", b: "e", c: "f" }]);
- * assertEquals(parse(string, { columns: ["x", "y", "z"] }), [{ x: "a", y: "b", z: "c" }, { x: "d", y: "e", z: "f" }]);
+ * const string = "a,b\nc,d,e";
+ * assertThrows(
+ *   () => parse(string, { fieldsPerRecord: 2 }),
+ *   SyntaxError,
+ *   "record on line 2: expected 2 fields but got 3",
+ * );
  * ```
  *
  * @typeParam T The options' type for parsing.
  * @param input The input to parse.
  * @param options The options for parsing.
- * @returns If you don't provide `options.skipFirstRow` and `options.columns`, it returns `string[][]`.
- *   If you provide `options.skipFirstRow` or `options.columns`, it returns `Record<string, unknown>[]`.
+ * @returns If you don't provide `options.skipFirstRow` or `options.columns`, it
+ * returns `string[][]`. If you provide `options.skipFirstRow` or
+ * `options.columns`, it returns `Record<string, string>[]`.
  */
 export function parse<const T extends ParseOptions>(
   input: string,
@@ -378,9 +520,9 @@ export function parse<const T extends ParseOptions>(
       headers = options.columns;
     }
 
-    const firstLineIndex = options.skipFirstRow ? 1 : 0;
+    const zeroBasedFirstLineIndex = options.skipFirstRow ? 1 : 0;
     return r.map((row, i) => {
-      return convertRowToObject(row, headers, firstLineIndex + i);
+      return convertRowToObject(row, headers, zeroBasedFirstLineIndex + i);
     }) as ParseResult<ParseOptions, T>;
   }
   return r as ParseResult<ParseOptions, T>;
diff --git a/csv/parse_stream.ts b/csv/parse_stream.ts
index 12ad67b7b5ec..5a9b9525081b 100644
--- a/csv/parse_stream.ts
+++ b/csv/parse_stream.ts
@@ -23,7 +23,7 @@ export interface CsvParseStreamOptions {
    * are ignored. With leading whitespace the comment character becomes part of
    * the field, even you provide `trimLeadingSpace: true`.
    *
-   * @default {"#"}
+   * By default, no character is considered to be a start of a comment.
    */
   comment?: string;
   /** Flag to trim the leading space of the value.
@@ -49,8 +49,8 @@ export interface CsvParseStreamOptions {
    * If negative, no check is made and records may have a variable number of
    * fields.
    *
-   * If the wrong number of fields is in a row, a {@linkcode ParseError} is
-   * thrown.
+   * If the wrong number of fields is in a row, a {@linkcode https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/SyntaxError | SyntaxError}
+   * is thrown.
    */
   fieldsPerRecord?: number;
   /**
@@ -108,17 +108,233 @@ export type RowType<T> = T extends undefined ? string[]
  * A `CsvParseStream` expects input conforming to
  * {@link https://www.rfc-editor.org/rfc/rfc4180.html | RFC 4180}.
  *
- * @example Usage
- * ```ts no-assert
+ * @example Usage with default options
+ * ```ts
  * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
  *
  * const source = ReadableStream.from([
- *   "name,age",
- *   "Alice,34",
- *   "Bob,24",
- *   "Charlie,45",
+ *   "name,age\n",
+ *   "Alice,34\n",
+ *   "Bob,24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream());
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   ["name", "age"],
+ *   ["Alice", "34"],
+ *   ["Bob", "24"],
+ * ]);
+ * assertType<IsExact<typeof result, string[][]>>(true);
+ * ```
+ *
+ * @example Skip first row with `skipFirstRow: true`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const source = ReadableStream.from([
+ *   "name,age\n",
+ *   "Alice,34\n",
+ *   "Bob,24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({ skipFirstRow: true }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   { name: "Alice", age: "34" },
+ *   { name: "Bob", age: "24" },
+ * ]);
+ * assertType<IsExact<typeof result, Record<string, string | undefined>[]>>(true);
+ * ```
+ *
+ * @example Specify columns with `columns` option
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const source = ReadableStream.from([
+ *   "Alice,34\n",
+ *   "Bob,24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   columns: ["name", "age"]
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   { name: "Alice", age: "34" },
+ *   { name: "Bob", age: "24" },
+ * ]);
+ * assertType<IsExact<typeof result, Record<"name" | "age", string>[]>>(true);
+ * ```
+ *
+ * @example Specify columns with `columns` option and skip first row with
+ * `skipFirstRow: true`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertType, IsExact } from "@std/testing/types"
+ *
+ * const source = ReadableStream.from([
+ *   "Alice,34\n",
+ *   "Bob,24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   columns: ["name", "age"],
+ *   skipFirstRow: true,
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [{ name: "Bob", age: "24" }]);
+ * assertType<IsExact<typeof result, Record<"name" | "age", string>[]>>(true);
+ * ```
+ *
+ * @example TSV (tab-separated values) with `separator: "\t"`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const source = ReadableStream.from([
+ *   "Alice\t34\n",
+ *   "Bob\t24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   separator: "\t",
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   ["Alice", "34"],
+ *   ["Bob", "24"],
+ * ]);
+ * ```
+ *
+ * @example Trim leading space with `trimLeadingSpace: true`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const source = ReadableStream.from([
+ *   "      Alice,34\n          ",
+ *   "Bob,     24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   trimLeadingSpace: true,
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   ["Alice", "34"],
+ *   ["Bob", "24"],
+ * ]);
+ * ```
+ *
+ * @example Quoted fields
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const source = ReadableStream.from([
+ *   `"a ""word""","com`,
+ *   `ma,","newline`,
+ *   `\n"\nfoo,bar,b`,
+ *   `az\n`,
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream());
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   ['a "word"', "comma,", "newline\n"],
+ *   ["foo", "bar", "baz"]
  * ]);
- * const parts = source.pipeThrough(new CsvParseStream());
+ * ```
+ *
+ * @example Allow lazy quotes with `lazyQuotes: true`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const source = ReadableStream.from([
+ *   `a "word","1"`,
+ *   `2",a","b`,
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   lazyQuotes: true,
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [['a "word"', '1"2', 'a"', 'b']]);
+ * ```
+ *
+ * @example Define comment prefix with `comment` option
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ *
+ * const source = ReadableStream.from([
+ *   "Alice,34\n",
+ *   "# THIS IS A COMMENT\n",
+ *   "Bob,24\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   comment: "#",
+ * }));
+ * const result = await Array.fromAsync(stream);
+ *
+ * assertEquals(result, [
+ *   ["Alice", "34"],
+ *   ["Bob", "24"],
+ * ]);
+ * ```
+ *
+ * @example Infer the number of fields from the first row with
+ * `fieldsPerRecord: 0`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertRejects } from "@std/assert/rejects";
+ *
+ * const source = ReadableStream.from([
+ *   "Alice,34\n",
+ *   "Bob,24,CA\n", // Note that this row has more fields than the first row
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   fieldsPerRecord: 0,
+ * }));
+ * const reader = stream.getReader();
+ * assertEquals(await reader.read(), { done: false, value: ["Alice", "34"] });
+ * await assertRejects(
+ *   () => reader.read(),
+ *   SyntaxError,
+ *   "record on line 2: expected 2 fields but got 3",
+ * );
+ * ```
+ *
+ * @example Enforce the number of field for each row with `fieldsPerRecord: 2`
+ * ```ts
+ * import { CsvParseStream } from "@std/csv/parse-stream";
+ * import { assertEquals } from "@std/assert/equals";
+ * import { assertRejects } from "@std/assert/rejects";
+ *
+ * const source = ReadableStream.from([
+ *   "Alice,34\n",
+ *   "Bob,24,CA\n",
+ * ]);
+ * const stream = source.pipeThrough(new CsvParseStream({
+ *   fieldsPerRecord: 2,
+ * }));
+ * const reader = stream.getReader();
+ * assertEquals(await reader.read(), { done: false, value: ["Alice", "34"] });
+ * await assertRejects(
+ *   () => reader.read(),
+ *   SyntaxError,
+ *   "record on line 2: expected 2 fields but got 3",
+ * );
  * ```
  *
  * @typeParam T The type of options for the stream.
@@ -258,20 +474,21 @@ export class CsvParseStream<
    * The instance's {@linkcode ReadableStream}.
    *
    * @example Usage
-   * ```ts no-assert
+   * ```ts
    * import { CsvParseStream } from "@std/csv/parse-stream";
+   * import { assertEquals } from "@std/assert/equals";
    *
    * const source = ReadableStream.from([
-   *   "name,age",
-   *   "Alice,34",
-   *   "Bob,24",
-   *   "Charlie,45",
+   *   "name,age\n",
+   *   "Alice,34\n",
+   *   "Bob,24\n",
    * ]);
-   * const parseStream = new CsvParseStream();
+   * const parseStream = new CsvParseStream({ skipFirstRow: true });
    * const parts = source.pipeTo(parseStream.writable);
-   * for await (const part of parseStream.readable) {
-   *   console.log(part);
-   * }
+   * assertEquals(await Array.fromAsync(parseStream.readable), [
+   *   { name: "Alice", age: "34" },
+   *   { name: "Bob", age: "24" },
+   * ]);
    * ```
    *
    * @returns The instance's {@linkcode ReadableStream}.
@@ -284,20 +501,21 @@ export class CsvParseStream<
    * The instance's {@linkcode WritableStream}.
    *
    * @example Usage
-   * ```ts no-assert
+   * ```ts
    * import { CsvParseStream } from "@std/csv/parse-stream";
+   * import { assertEquals } from "@std/assert/equals";
    *
    * const source = ReadableStream.from([
-   *   "name,age",
-   *   "Alice,34",
-   *   "Bob,24",
-   *   "Charlie,45",
+   *   "name,age\n",
+   *   "Alice,34\n",
+   *   "Bob,24\n",
    * ]);
-   * const parseStream = new CsvParseStream();
+   * const parseStream = new CsvParseStream({ skipFirstRow: true });
    * const parts = source.pipeTo(parseStream.writable);
-   * for await (const part of parseStream.readable) {
-   *   console.log(part);
-   * }
+   * assertEquals(await Array.fromAsync(parseStream.readable), [
+   *   { name: "Alice", age: "34" },
+   *   { name: "Bob", age: "24" },
+   * ]);
    * ```
    *
    * @returns The instance's {@linkcode WritableStream}.
diff --git a/csv/parse_stream_test.ts b/csv/parse_stream_test.ts
index 62fa5f742f9e..6288ec4ff546 100644
--- a/csv/parse_stream_test.ts
+++ b/csv/parse_stream_test.ts
@@ -350,14 +350,23 @@ x,,,
         columns: ["foo", "bar", "baz"],
       },
       {
-        name: "mismatching number of headers and fields",
+        name: "mismatching number of headers and fields 1",
         input: "a,b,c\nd,e",
         skipFirstRow: true,
         columns: ["foo", "bar", "baz"],
         error: {
           klass: Error,
-          msg:
-            "Error number of fields line: 1\nNumber of fields found: 3\nExpected number of fields: 2",
+          msg: "record on line 2 has 2 fields, but the header has 3 fields",
+        },
+      },
+      {
+        name: "mismatching number of headers and fields 2",
+        input: "a,b,c\nd,e,,g",
+        skipFirstRow: true,
+        columns: ["foo", "bar", "baz"],
+        error: {
+          klass: Error,
+          msg: "record on line 2 has 4 fields, but the header has 3 fields",
         },
       },
       {
diff --git a/csv/parse_test.ts b/csv/parse_test.ts
index e2b3c74edb13..f5dae7abca5b 100644
--- a/csv/parse_test.ts
+++ b/csv/parse_test.ts
@@ -23,6 +23,7 @@ Deno.test({
         );
       },
     });
+
     await t.step({
       name: "CRLF",
       fn() {
@@ -97,6 +98,42 @@ Deno.test({
       },
     });
 
+    await t.step({
+      name: "BlankField",
+      fn() {
+        const input = "a,b,c\nd,,f";
+        assertEquals(
+          parse(input),
+          [["a", "b", "c"], ["d", "", "f"]],
+        );
+      },
+    });
+
+    await t.step({
+      name: "BlankField2",
+      fn() {
+        const input = "a,b,c\nd,,f";
+        assertEquals(
+          parse(input, { skipFirstRow: true }),
+          [{ a: "d", b: "", c: "f" }],
+        );
+      },
+    });
+
+    await t.step({
+      name: "BlankField3",
+      fn() {
+        const input = "a,b,c\nd,,f";
+        assertEquals(
+          parse(input, { columns: ["one", "two", "three"] }),
+          [
+            { one: "a", two: "b", three: "c" },
+            { one: "d", two: "", three: "f" },
+          ],
+        );
+      },
+    });
+
     await t.step({
       name: "BlankLine",
       fn() {
@@ -257,7 +294,7 @@ Deno.test({
         assertThrows(
           () => parse(input, { fieldsPerRecord: 0 }),
           SyntaxError,
-          "record on line 2: wrong number of fields",
+          "record on line 2: expected 3 fields but got 2",
         );
       },
     });
@@ -268,10 +305,21 @@ Deno.test({
         assertThrows(
           () => parse(input, { fieldsPerRecord: 2 }),
           SyntaxError,
-          "record on line 1: wrong number of fields",
+          "record on line 1: expected 2 fields but got 3",
         );
       },
     });
+    await t.step({
+      name: "NegativeFieldsPerRecord",
+      fn() {
+        const input = `a,b,c\nd,e`;
+        const output = [
+          ["a", "b", "c"],
+          ["d", "e"],
+        ];
+        assertEquals(parse(input, { fieldsPerRecord: -1 }), output);
+      },
+    });
     await t.step({
       name: "FieldCount",
       fn() {
@@ -783,7 +831,7 @@ c"d,e`;
       },
     });
     await t.step({
-      name: "mismatching number of headers and fields",
+      name: "mismatching number of headers and fields 1",
       fn() {
         const input = "a,b,c\nd,e";
         assertThrows(
@@ -793,7 +841,22 @@ c"d,e`;
               columns: ["foo", "bar", "baz"],
             }),
           Error,
-          "Error number of fields line: 1\nNumber of fields found: 3\nExpected number of fields: 2",
+          "record on line 2 has 2 fields, but the header has 3 fields",
+        );
+      },
+    });
+    await t.step({
+      name: "mismatching number of headers and fields 2",
+      fn() {
+        const input = "a,b,c\nd,e,,g";
+        assertThrows(
+          () =>
+            parse(input, {
+              skipFirstRow: true,
+              columns: ["foo", "bar", "baz"],
+            }),
+          Error,
+          "record on line 2 has 4 fields, but the header has 3 fields",
         );
       },
     });