How to Import CSV to BigQuery (4 Methods)

BigQuery handles CSV imports differently from traditional databases. There's no COPY command or server-side file access. Instead, you load files through the BigQuery API -- either from your local machine or from Google Cloud Storage (GCS). Here are the practical methods.

Method 1: bq CLI

The bq command-line tool is the quickest way to load a local CSV:

bq load \
  --source_format=CSV \
  --autodetect \
  --skip_leading_rows=1 \
  mydataset.users \
  ./users.csv

Key flags:

• --autodetect -- BigQuery infers the schema from the file
• --skip_leading_rows=1 -- skips the header row
• --replace -- overwrites the table (default is append)
• --max_bad_records=10 -- allows up to 10 malformed rows before failing

To specify the schema explicitly instead of auto-detecting:

bq load \
  --source_format=CSV \
  --skip_leading_rows=1 \
  mydataset.users \
  ./users.csv \
  name:STRING,email:STRING,age:INTEGER,signup_date:DATE

From Google Cloud Storage:

bq load \
  --source_format=CSV \
  --autodetect \
  --skip_leading_rows=1 \
  mydataset.users \
  gs://my-bucket/users.csv

Loading from GCS is faster for large files and supports wildcards (gs://my-bucket/users_*.csv).

Method 2: Google Cloud Console

The BigQuery web UI provides a visual import:

1. Open BigQuery in the Google Cloud Console
2. Select your dataset in the explorer panel
3. Click Create Table
4. Set source to Upload and select your CSV file
5. BigQuery shows schema auto-detection -- review and adjust
6. Configure table name and write disposition (append, overwrite, or create if empty)
7. Click Create Table

The console has a 100 MB file size limit for direct uploads. For larger files, upload to GCS first, then load from there.

Method 3: Python Client Library

The google-cloud-bigquery Python library provides programmatic control:

from google.cloud import bigquery
 
client = bigquery.Client()
 
job_config = bigquery.LoadJobConfig(
    source_format=bigquery.SourceFormat.CSV,
    skip_leading_rows=1,
    autodetect=True,
    write_disposition=bigquery.WriteDisposition.WRITE_TRUNCATE,
)
 
table_id = "my-project.mydataset.users"
 
with open("users.csv", "rb") as f:
    job = client.load_table_from_file(f, table_id, job_config=job_config)
 
job.result()  # Wait for completion
print(f"Loaded {job.output_rows} rows")

For explicit schema control:

job_config = bigquery.LoadJobConfig(
    schema=[
        bigquery.SchemaField("name", "STRING"),
        bigquery.SchemaField("email", "STRING"),
        bigquery.SchemaField("age", "INTEGER"),
        bigquery.SchemaField("signup_date", "DATE"),
    ],
    skip_leading_rows=1,
    source_format=bigquery.SourceFormat.CSV,
)

Method 4: Cloud Storage + Scheduled Loads

For recurring imports, upload files to a GCS bucket and configure a scheduled load or trigger:

# Upload to GCS
gsutil cp users.csv gs://my-bucket/imports/users.csv
 
# Load into BigQuery
bq load --source_format=CSV --autodetect --skip_leading_rows=1 \
  mydataset.users gs://my-bucket/imports/users.csv

For automated pipelines, use BigQuery Data Transfer Service or set up a Cloud Function triggered by GCS object creation.

Common Gotchas

Auto-detect isn't always right. BigQuery's schema auto-detection samples a subset of rows. If the first 100 rows of a column are integers but row 101 has a string, the load will fail. When in doubt, define the schema explicitly.

Quoted newlines. BigQuery's CSV parser handles quoted fields containing newlines, but you may need to set --allow_quoted_newlines explicitly:

bq load --source_format=CSV --allow_quoted_newlines \
  mydataset.users ./users.csv

NULL handling. BigQuery treats empty strings and the literal string "" (two double quotes) differently. An empty field in CSV becomes NULL. A field containing "" becomes an empty string. This catches people who expect empty strings to be stored as empty strings.

Date and timestamp formats. BigQuery accepts dates as YYYY-MM-DD and timestamps as YYYY-MM-DD HH:MM:SS[.ffffff] [timezone]. Other formats will fail. Preprocess your CSV or load as STRING and convert with PARSE_DATE or PARSE_TIMESTAMP in SQL.

File size limits. Direct upload via bq load or the console supports up to 5 TB per load job. The console UI limits uploads to 100 MB. For large local files, upload to GCS first.

Costs. Loading data into BigQuery is free. Storage costs apply once data is loaded (about $0.02/GB/month for active storage as of 2026). Queries against the data are billed by bytes scanned.

Mako connects to PostgreSQL, MySQL, MongoDB, BigQuery, Snowflake, and ClickHouse with AI-powered autocomplete. Try it free at mako.ai.