The Abbreviated Math Anxiety Scale (AMAS) is commonly used to compare groups on math anxiety. Group comparisons should however be preceded by a demonstration of metric and scalar measurement invariance, which is currently only available for undergraduate students in the USA. This study tested for metric and scalar measurement invariance of AMAS scores across sexes and age groups and investigated reliability and validity evidence. Dutch secondary school students (N = 1504) and university students (N = 629) completed the AMAS and reported their math grades. Subsamples also completed questionnaires on related (e.g., trait anxiety) and unrelated constructs (e.g., anxiety for learning a foreign language). Results of multi-group Confirmatory Factor Analyses showed the 2-factor structure that was detected in earlier studies and supported partial metric and scalar invariance, although cross-loadings for one item were included for the female university students group to improve model fit. Reliability ranged from acceptable to good and validity was supported.