CONTEXT: Recommendations conflict regarding universal application of formal screening instruments in primary care (PC) and PC-like settings for autism spectrum disorder (ASD).

OBJECTIVES: We systematically reviewed evidence for universal screening of children for ASD in PC.

DATA SOURCES: We searched Medline, PsychInfo, Educational Resources Informational Clearinghouse, and Cumulative Index of Nursing and Allied Health Literature.

STUDY SELECTION: We included studies in which researchers report psychometric properties of screening tools in unselected populations across PC and PC-like settings.

DATA EXTRACTION: At least 2 authors reviewed each study, extracted data, checked accuracy, and assigned quality ratings using predefined criteria.

RESULTS: We found evidence for moderate to high positive predictive values for ASD screening tools to identify children aged 16 to 40 months and 1 study for ≥48 months in PC and PC-like settings. Limited evidence evaluating sensitivity, specificity, and negative predictive value of instruments was available. No studies directly evaluated the impact of screening on treatment or harm.

LIMITATIONS: Potential limitations include publication bias, selective reporting within studies, and a constrained search.

CONCLUSIONS: ASD screening tools can be used to accurately identify percentages of unselected populations of young children for ASD in PC and PC-like settings. The scope of challenges associated with establishing direct linkage suggests that clinical and policy groups will likely continue to guide screening practices. ASD is a common neurodevelopmental disorder associated with significant life span costs. Growing evidence supports functional gains and improved outcomes for young children receiving intensive intervention, so early identification on a population level is a pressing public health challenge.