Standardized language testing may underestimate abilities of young children with autism due to lack of motivation and social or behavioral difficulties in the testing environment. This study examined how standardized measures capture functional language used in the natural environment by young children with autism. The results of a standard pictorial naming assessment were compared to the results of a 10-minute communication probe with the primary parent for 34 young children with autism (average age at pre-intervention = 29 mos). At three timepoints (prior to intervention, after 200 hours of intervention, and six months after termination), a higher percentage of children had expressive language on the naturalistic assessment as compared to the standardized measure. The naturalistic assessment captured 100% of children with words on the standardized measure, but the standardized measure did not capture all of the children with words on the naturalistic assessment. This study suggests that supplementing standardized measures with a 10-minute naturalistic assessment adds valuable information about linguistic abilities in young children with autism.