Procedures to evaluate the quality of musical performance, like psychological tests generally, vary in reliability and validity. How can both be optimised? The subjective world of jurors comprises input (sensations) and output (thoughts and emotions). Cognitive approaches to performance evaluation, in which different aspects are analytically considered and the results combined, are more reliable than emotional approaches, which are ultimately based on holistic ‘gut reactions’. However, emotional approaches may be more valid. Both depend on the serial order in which performances are presented to evaluators, suggesting a need for independent, computer-controlled procedures in which jurors evaluate performances in different random orders. Jurors can be influenced by performers’ appearance and movements, as well as knowledge about past performances, whether or not they believe they should be; additional blind evaluations (sound only) could help. Evaluations may depend primarily on specific features such as right-hand melodic phrasing in romantic piano music. It would be interesting to systematically track the emotional state of jurors during a competition in order to better understand the interaction between their thoughts and emotions. In general, traditional approaches to performance evaluation might be supplemented (not replaced) by psychologically inspired, computer-based procedures. A ‘superjury’ could compare evaluations from different methods or ‘subjuries’.