The objective of the study was to compare the performance of four short‐form versions of the Child Perceptions Questionnaire (CPQ~11–14~) with that of the long‐form version in a random population sample of 12‐ and 13‐yr‐old children from New Zealand in order to determine which short‐form version was the most valid. Children (n = 430, participation rate 74.1%) completed the 37‐item CPQ~11–14~. Two separate 8‐ and 16‐item short‐form versions were previously developed using (a) item impact and (b) regression methods. The four different short‐form scales were compared with the full CPQ~11–14~ on their construct validity. The children were examined for malocclusion (using the Dental Aesthetic Index) and for dental caries by a single examiner (L.F.P.). All short‐form versions revealed substantial variability in overall oral health‐related quality of life (OHRQoL). Cronbach’s alpha ranged from 0.73 (Regression Short Form [RSF]‐8) to 0.86 (RSF‐16). For all short‐form versions, mean scores were positively associated with self‐rated oral health and overall wellbeing; associations with the latter were stronger. All short‐form versions detected OHRQoL gradients, as hypothesized, across ascending categories of caries and malocclusion. These findings suggest that the short‐form versions of the CPQ~11–14~ all show acceptable properties, but that the 16‐item versions perform better (and are essentially equivalent); however, the stronger theoretical underpinning of the item‐impact‐derived 16‐item short‐form version suggests that it shows the most promise.