This paper discusses the conditions favouring transfer of laboratory-based science process skills. It examines the generalisability or otherwise of these skills for students carrying out whole investigations, using a sample of 135 eighth graders in Singapore. The evidence for generalisability across different investigations is elicited in three ways: (1) pair-wise comparison of performance across different investigations for each science process skill, using Pearson's tetrachoric correlations as the case may be; (2) agreement in ranking of performance across investigations for each of the science process skilb, using Kendall's Coefficient of Concordance, W; (3) group mean comparison across investigations for each of the science process skilb, using t-tests.