Constraints are a major focus of theoretical and empirical study in leisure research.  However, analyses of psychometric properties of constraints measures are rare. This study assessed the factor structure and invariance of a leisure constraints measure applied to sport participation among middle school adolescents (ages 11 to 15 years old) using panel data over two time periods nine months apart (N=2,029).We compared the validity of three baseline models: a theoretical 3-factor model, a 7-factor model, and a second-order factor model. The 7-factor model exhibited a more appropriate measurement model based on fit indices and statistical comparisons. Using the 7-factor model, construct validity and invariance of the model over time (Time 1 and Time 2) was examined by gender, grade level, race/ethnicity, and socioeconomic status. The model was found to be invariant at the factor loading and intercept levels, while there was some evidence of non-invariance at further constrained levels. In general, the 7-factor model appeared to be psychometrically stable and applicable over a range of demographic subgroups, lending evidence to its use as a measure in leisure or  sport participation constraint studies that involve socio-demographic comparisons. The findings suggest that while the theoretical structure may be effective in simplifying constraint categories, a more detailed specification using subfactors of the dimensions is better for measurement.