Abstract: Background: Our previous research (Pediatrics 2010:126) found a strong association between caregiver oral health literacy (OHL) and children’s oral health status; however, we found a weak association with oral health behaviors (OHBs). We hypothesize that this may be due to social desirability bias (SDB). Our objectives were to compare caregivers’ responses to traditional OHB items and newer SDB-modulating items, and to examine the association of caregiver literacy with OHBs. Methods: We performed a cross-sectional study of 102 caregiver-child dyads, collecting data for OHBs using both traditional and new SDB-modulating items. We measured OHL using REALD-30, a validated word recognition test. We relied upon percent agreement and Cohen’s kappa (k) to quantify the concordance in caregivers’ responses and multivariate log-binomial regression to estimate the impact of OHL on OHBs. Results: Caregivers’ mean REALD-30 score was 20.7 (SD = 6.0), range 1-30. We found an association between OHL and 4 of 8 OHBs examined. A subset of behavior questions compared traditional versus SDB-modulating items: history of bottle-feeding: agreement = 95%, k = 0.83 (95% CL:0.68,0.99); daily tooth brushing: agreement = 78%, k = 0.25 (95% CL:0.04,0.46); fluoridated toothpaste use: agreement = 88%, k = 0.67 (95% CL:0.49,0.85). After controlling for caregivers’ race, marital status and study site, higher literacy scores remained associated with a decreased prevalence of parental report of “decided not brush the child’s teeth because it would be frustrating”. Conclusions: Agreement between responses was high for 2 of 3 behavior items. Item 3 (tooth brushing frequency) revealed discordance, likely due to SDB. Use of the SDB-modulating items appears to yield a better estimate of OHB.