diff --git a/bigquery/external.go b/bigquery/external.go
index 3f8647e568a..36a1ab64ac3 100644
--- a/bigquery/external.go
+++ b/bigquery/external.go
@@ -100,19 +100,17 @@ type ExternalDataConfig struct {
 
 func (e *ExternalDataConfig) toBQ() bq.ExternalDataConfiguration {
 	q := bq.ExternalDataConfiguration{
-		SourceFormat:        string(e.SourceFormat),
-		SourceUris:          e.SourceURIs,
-		Autodetect:          e.AutoDetect,
-		Compression:         string(e.Compression),
-		IgnoreUnknownValues: e.IgnoreUnknownValues,
-		MaxBadRecords:       e.MaxBadRecords,
+		SourceFormat:            string(e.SourceFormat),
+		SourceUris:              e.SourceURIs,
+		Autodetect:              e.AutoDetect,
+		Compression:             string(e.Compression),
+		IgnoreUnknownValues:     e.IgnoreUnknownValues,
+		MaxBadRecords:           e.MaxBadRecords,
+		HivePartitioningOptions: e.HivePartitioningOptions.toBQ(),
 	}
 	if e.Schema != nil {
 		q.Schema = e.Schema.toBQ()
 	}
-	if e.HivePartitioningOptions != nil {
-		q.HivePartitioningOptions = e.HivePartitioningOptions.toBQ()
-	}
 	if e.Options != nil {
 		e.Options.populateExternalDataConfig(&q)
 	}
@@ -121,13 +119,14 @@ func (e *ExternalDataConfig) toBQ() bq.ExternalDataConfiguration {
 
 func bqToExternalDataConfig(q *bq.ExternalDataConfiguration) (*ExternalDataConfig, error) {
 	e := &ExternalDataConfig{
-		SourceFormat:        DataFormat(q.SourceFormat),
-		SourceURIs:          q.SourceUris,
-		AutoDetect:          q.Autodetect,
-		Compression:         Compression(q.Compression),
-		IgnoreUnknownValues: q.IgnoreUnknownValues,
-		MaxBadRecords:       q.MaxBadRecords,
-		Schema:              bqToSchema(q.Schema),
+		SourceFormat:            DataFormat(q.SourceFormat),
+		SourceURIs:              q.SourceUris,
+		AutoDetect:              q.Autodetect,
+		Compression:             Compression(q.Compression),
+		IgnoreUnknownValues:     q.IgnoreUnknownValues,
+		MaxBadRecords:           q.MaxBadRecords,
+		Schema:                  bqToSchema(q.Schema),
+		HivePartitioningOptions: bqToHivePartitioningOptions(q.HivePartitioningOptions),
 	}
 	switch {
 	case q.CsvOptions != nil:
@@ -141,9 +140,6 @@ func bqToExternalDataConfig(q *bq.ExternalDataConfiguration) (*ExternalDataConfi
 			return nil, err
 		}
 	}
-	if q.HivePartitioningOptions != nil {
-		e.HivePartitioningOptions = bqToHivePartitioningOptions(q.HivePartitioningOptions)
-	}
 	return e, nil
 }
 
diff --git a/bigquery/load.go b/bigquery/load.go
index 9fa0190685b..1b2d3c6bfa7 100644
--- a/bigquery/load.go
+++ b/bigquery/load.go
@@ -65,6 +65,10 @@ type LoadConfig struct {
 	// For ingestion from datastore backups, ProjectionFields governs which fields
 	// are projected from the backup.  The default behavior projects all fields.
 	ProjectionFields []string
+
+	// HivePartitioningOptions allows use of Hive partitioning based on the
+	// layout of objects in Cloud Storage.
+	HivePartitioningOptions *HivePartitioningOptions
 }
 
 func (l *LoadConfig) toBQ() (*bq.JobConfiguration, io.Reader) {
@@ -81,6 +85,7 @@ func (l *LoadConfig) toBQ() (*bq.JobConfiguration, io.Reader) {
 			SchemaUpdateOptions:                l.SchemaUpdateOptions,
 			UseAvroLogicalTypes:                l.UseAvroLogicalTypes,
 			ProjectionFields:                   l.ProjectionFields,
+			HivePartitioningOptions:            l.HivePartitioningOptions.toBQ(),
 		},
 	}
 	media := l.Src.populateLoadConfig(config.Load)
@@ -100,6 +105,7 @@ func bqToLoadConfig(q *bq.JobConfiguration, c *Client) *LoadConfig {
 		SchemaUpdateOptions:         q.Load.SchemaUpdateOptions,
 		UseAvroLogicalTypes:         q.Load.UseAvroLogicalTypes,
 		ProjectionFields:            q.Load.ProjectionFields,
+		HivePartitioningOptions:     bqToHivePartitioningOptions(q.Load.HivePartitioningOptions),
 	}
 	var fc *FileConfig
 	if len(q.Load.SourceUris) == 0 {
diff --git a/bigquery/load_test.go b/bigquery/load_test.go
index eb9ec38804d..ca3fe9da75f 100644
--- a/bigquery/load_test.go
+++ b/bigquery/load_test.go
@@ -342,6 +342,31 @@ func TestLoad(t *testing.T) {
 				return j
 			}(),
 		},
+		{
+			dst: c.Dataset("dataset-id").Table("table-id"),
+			src: func() *GCSReference {
+				g := NewGCSReference("uri")
+				g.SourceFormat = Parquet
+				return g
+			}(),
+			config: LoadConfig{
+				HivePartitioningOptions: &HivePartitioningOptions{
+					Mode:                   CustomHivePartitioningMode,
+					SourceURIPrefix:        "source_uri",
+					RequirePartitionFilter: true,
+				},
+			},
+			want: func() *bq.Job {
+				j := defaultLoadJob()
+				j.Configuration.Load.SourceFormat = "PARQUET"
+				j.Configuration.Load.HivePartitioningOptions = &bq.HivePartitioningOptions{
+					Mode:                   "CUSTOM",
+					RequirePartitionFilter: true,
+					SourceUriPrefix:        "source_uri",
+				}
+				return j
+			}(),
+		},
 	}
 
 	for i, tc := range testCases {