Vitamin D deficiency has adverse health effects in young children. Our aims were to determine predictors of vitamin D status and then to use these factors to develop a practical tool to predict low 25(OH)D concentrations in preschool New Zealand children. A cross-sectional sample of 1329 children aged 2 to <5 years were enrolled from throughout New Zealand in late-winter to spring 2012. 25-Hydroxyvitamin D (25(OH)D) was measured on dried blood spot (DBS) samples collected using finger-prick sampling. Caregivers completed a questionnaire. Mean (SD) DBS 25(OH)D concentration was 52(19)nmol/L. 25(OH)D < 25 nmol/L was present in 86(7%), 25(OH)D < 50 nmol/L in 642(48%), 25(OH)D 50- < 75 nmol/L in 541(41%) and 25(OH)D > 75 nmol/L in 146(11%) of children. Factors independently associated with the risk of 25(OH)D < 25 nmol/L were female gender (OR 1.92,95%CI 1.17-3.14), other non-European ethnicities (not including Māori or Pacific) (3.51,1.89-6.50), had olive-dark skin colour (4.52,2.22-9.16), did not take vitamin D supplements (2.56,1.06-6.18), had mothers with less than secondary-school qualifications (5.00,2.44-10.21) and lived in more deprived households (1.27,1.06-1.53). Children who drank toddler milk (vitamin D fortified cow's milk formula marketed to young children) had a zero risk of 25(OH)D < 25 nmol/L. The predictive tool identified children at risk of 25(OH)D < 25 nmol/L with sensitivity 42%, specificity 97% and ROC area-under-curve 0.76(95%CI 0.67-0.86, p < 0.001). Predictors of low vitamin D status were consistent with those identified in previous studies of New Zealand children. The tool had insufficient predictive ability for use in clinical situations, and suggests a need to promote safe, inexpensive testing to determine vitamin D status in preschool children.