Selective eating in children is commonly measured by parental report questionnaires, yet it is unknown if parents accurately estimate their child’s selective eating behavior. The objectives of this study were to test the validity and stability of two measures of selective eating using observed child behavior. Low-income mother-child dyads participated in a videotaped laboratory eating protocol at two time points (baseline: mean child age = 5.9 years; follow-up: mean child age = 8.6 years), during which they were presented with a familiar and an unfamiliar vegetable. Videos were reliably coded for child selective eating behaviors: amount consumed, child hedonic rating of vegetables, child compliance with maternal prompts to eat, latency to first bite, number of bites, and negative utterances. Mothers completed the Child Eating Behavior Questionnaire Food Fussiness (CEBQ FF) scale and the Food Neophobia Scale (FNS) at both time points. Questionnaire validity, test re-test reliability of measured behaviors, and discriminant validity of questionnaires were examined in the full sample. CEBQ FF scores and FNS scores were both inversely correlated with the quantity consumed, child hedonic rating, and compliance with prompts to eat for both familiar and unfamiliar vegetables at baseline and at follow up. CEBQ FF and FNS scores were inversely correlated with number of bites (for both foods), positively correlated with latency to first bite (for both foods), and inversely correlated with child negative utterances (for the familiar food only). Notably, FNS scores correlated with observed behavior for both familiar and unfamiliar foods, rather than demonstrating a specific association with unfamiliar foods only. This study supports the validity of the CEBQ FF and FNS in low-income early school-aged children.