Objective To develop a reliable self-report tool for measuring child booster seat use among Latino families.

Design Cross-sectional and observational survey of a convenience sample.

Setting Five retail stores in King County, Washington.

Participants 50 parents of children 4–8 years old that self-identified as Latino or Hispanic.

Main exposures Parent-reported measures of how often the child uses a booster seat, if the child used a booster seat on the last trip, how often the child complains about using a booster seat, how often the child asks to not use a booster seat, and how often other families they know use a booster seat.

Outcome measure Observed booster seat use by child.

Results 26 children (52%) were observed using a booster seat. Parent-reported booster seat use was a poor predictor of observed booster seat use. Although 34 parents reported that their child ‘always’ uses a booster seat, 8 (24%) of these children were not using a booster seat. A logistic model to predict booster seat use had a sensitivity of 81% and a specificity of 71%, and misclassified 24% of the participants' observed use.

Conclusions Reliance on parent-reported booster seat use significantly overstated observed booster seat use in the study. Among this study population, accurate determination of booster seat use required direct observation.