Summary.Science teachers have expressed dissatisfaction regarding both the validity an the predictive value of the conventional practical tests in science. The proposed new secondary school examinations have necessitated a rethinking of the purpose of such tests, and the investigation here described is a contribution to this field. It has been carried out as far as possible within the kind of conditions which may obtain for these new examinations.Three forms of practical science tests were based on six hypothetical components of practical ability in science. Tests were constructed in biology, chemistry and physics. Using 196 secondary school pupils, the results from these tests were compared with scores on tests of attainment and ability in science, and with corresponding teachers' estimates. The analysis of these scores showed wide differences between schools and between identical test patterns in the three science subjects. The differences between schools were no doubt temporary facets of a new and disturbing situation. Greater theroetical interest attaches to the differences between the test adaptability of the separate subjects, and it would seem that identical test procedures cannot be used in each field.