Concerns about the value of physical testing and apparently declining test performance in junior basketball players prompted this retrospective study of trends in anthropometric and fitness test scores related to recruitment age and recruitment year. The participants were 1011 females and 1087 males entering Basketball Australia's State and National programmes (1862 and 236 players, respectively). Players were tested on 2.6 +/- 2.0 (mean +/- s) occasions over 0.8 +/- 1.0 year. Test scores were adjusted to recruitment age (14-19 years) and recruitment year (1996-2003) using mixed modelling. Effects were estimated by log transformation and expressed as standardized (Cohen) differences in means. National players scored more favourably than State players on all tests, with the differences being generally small (standardized differences, 0.2-0.6) or moderate (0.6-1.2). On all tests, males scored more favourably than females, with large standardized differences (>1.2). Athletes entering at age 16 performed at least moderately better than athletes entering at age 14 on most tests (standardized differences, 0.7-2.1), but test scores often plateaued or began to deteriorate at around 17 years. Some fitness scores deteriorated over the 8-year period, most notably a moderate increase in sprint time and moderate (National male) to large (National female) declines in shuttle run performance. Variation in test scores between National players was generally less than that between State players (ratio of standard deviations, 0.83-1.18). More favourable means and lower variability in athletes of a higher standard highlight the potential utility of these tests in junior basketball programmes, although secular declines should be a major concern of Australian basketball coaches.