Skip to main content

​
ZodError Structure

When validation fails, Zod throws a ZodError that contains detailed information about all validation issues.

​
Basic Error Structure

import * as z from 'zod';

const schema = z.object({
  f1: z.number(),
  f3: z.string().nullable(),
  f4: z.array(z.object({ t: z.union([z.string(), z.boolean()]) }))
});

const result = schema.safeParse({});

if (!result.success) {
  console.log(result.error);
  // ZodError with issues array
}
The ZodError object contains:
  • issues: Array of $ZodIssue objects describing each validation failure
  • message: JSON stringified representation of all issues
  • name: Always "ZodError"

​
Issue Types

Zod provides different issue types for various validation failures:

​
Invalid Type Issue

interface $ZodIssueInvalidType {
  readonly code: "invalid_type";
  readonly expected: "string" | "number" | "boolean" | "array" | "object" | ...;
  readonly input?: unknown;
  readonly path: PropertyKey[];
  readonly message: string;
}
Example: 
const result = z.string().safeParse(234);
// {
//   "expected": "string",
//   "code": "invalid_type",
//   "path": [],
//   "message": "Invalid input: expected string, received number"
// }

​
Too Small/Too Big Issues

const result = z.array(z.string()).min(3).safeParse(["asdf", "qwer"]);

if (!result.success) {
  console.log(result.error.issues[0]);
  // {
  //   "origin": "array",
  //   "code": "too_small",
  //   "minimum": 3,
  //   "inclusive": true,
  //   "path": [],
  //   "message": "Too small: expected array to have >=3 items"
  // }
}

​
Invalid Format Issue

const schema = z.string().email();
const result = schema.safeParse("asdfsdf");

// Issue: { code: "invalid_format", format: "email", ... }

​
Custom Issue

const schema = z.number().refine((val) => val > 3, {
  message: "override"
});

const result = schema.safeParse(2);
// {
//   "code": "custom",
//   "path": [],
//   "message": "override"
// }

​
Error Formatting

Zod provides multiple ways to format errors for different use cases.

​
Flattened Errors

The flatten() method groups errors into form-level and field-level errors: 
const Test = z.object({
  f1: z.number(),
  f2: z.string().optional(),
  f3: z.string().nullable(),
  f4: z.array(z.object({ t: z.union([z.string(), z.boolean()]) }))
});

const result = Test.safeParse({});
const flattened = result.error!.flatten();

console.log(flattened);
// {
//   "fieldErrors": {
//     "f1": ["Invalid input: expected number, received undefined"],
//     "f3": ["Invalid input: expected string, received undefined"],
//     "f4": ["Invalid input: expected array, received undefined"]
//   },
//   "formErrors": []
// }

​
Custom Flatten Mapper

You can provide a custom mapper function to transform issues: 
type ErrorType = { message: string; code: number };

const flattened = result.error!.flatten((iss) => ({
  message: iss.message,
  code: 1234
}));

// Returns ErrorType[] instead of string[]

​
Formatted Errors (Tree Structure)

The format() method creates a nested tree structure matching your schema: 
const schema = z.object({
  inner: z.object({
    name: z.string().refine((val) => val.length > 5).array()
      .refine((val) => val.length <= 1)
  })
});

const result = schema.safeParse({
  inner: { name: ["aasd", "asdfasdfasfd"] }
});

const formatted = result.error!.format();
// {
//   "_errors": [],
//   "inner": {
//     "_errors": [],
//     "name": {
//       "_errors": ["Invalid input"],
//       // ... nested errors
//     }
//   }
// }

​
Root Level Formatting

const schema = z.string().email();
const result = schema.safeParse("asdfsdf");

const formatted = result.error!.format();
// {
//   "_errors": ["Invalid email address"]
// }

​
Custom Error Messages

​
String Parameter

The simplest way to customize error messages: 
// For schema constructors
const schema = z.string("Bad!");
schema.safeParse(123).error!.issues[0].message; // "Bad!"

// For refinements
const refined = z.number().refine((x) => x > 3, "override");
refined.safeParse(2).error!.issues[0].message; // "override"

// For checks
const minLength = z.string().min(5, "Too short!");
minLength.safeParse("abc").error!.issues[0].message; // "Too short!"

​
Message Object

const schema = z.string().min(10, { message: "override" });

const result = schema.safeParse("tooshort");
// message: "override"

​
Refinement with Custom Path

const schema = z.object({
  password: z.string(),
  confirm: z.string()
}).refine((val) => val.confirm === val.password, {
  path: ["confirm"],
  message: "Passwords don't match"
});

const result = schema.safeParse({
  password: "peanuts",
  confirm: "qeanuts"
});

const error = result.error!.format();
// error.confirm._errors = ["Passwords don't match"]

​
Refinement with Params

const schema = z.number().refine((val) => val >= 3, {
  params: { minimum: 3 }
});

const result = schema.safeParse(2);
// {
//   "code": "custom",
//   "path": [],
//   "params": { "minimum": 3 },
//   "message": "Invalid input"
// }
Use params to pass additional metadata to error maps for dynamic message generation.

​
Error Utilities

​
Adding Issues

const err = new z.ZodError([]);

err.addIssue({
  code: "invalid_type",
  expected: "object",
  path: [],
  message: "Expected object",
  input: "adf"
});

// Or add multiple issues
err.addIssues([issue1, issue2]);

​
Checking Empty Errors

if (err.isEmpty) {
  // No validation errors
}
Don’t call .parse() on schemas with async refinements. It will throw an error. Use .parseAsync() instead.

​
TypeScript Type Inference

type FormattedError = z.inferFormattedError<typeof schema>;
type FlattenedError = z.inferFlattenedErrors<typeof schema>;

// With custom mapper
type CustomFormatted = z.inferFormattedError<typeof schema, number>;

​
Best Practices

  1. Use safeParse in production - Returns a result object instead of throwing
  2. Flatten for forms - Use .flatten() for form validation to map errors to fields
  3. Format for nested data - Use .format() for complex nested objects
  4. Add context with params - Include params in refinements for dynamic error messages
  5. Set custom paths - Use the path option in refinements to target specific fields

Build docs developers (and LLMs) love

Get started for freeTalk to us