Instructional animation research has been extensive but the results are inconsistent. Amongst a number of possible factors to explain these inconclusive results (e.g., the negative influence of transient information), the influence of spatial ability and gender are less explored. This paper reports three experiments that compared the effectiveness of learning a hand-manipulative task (Lego construction) under various conditions with direct examination of the relationship between gender, spatial ability and instructional visualisation. Regression analyses revealed that only one objective measure related to spatial ability (Corsi test) predicted overall test performance, whereas the Card Rotations Test and the Mental Rotations Test did not. However, there was a number of significant gender spatial ability interactions showing that the spatial ability predictors of male performance were different from those of females. Furthermore a number of subjective measures of spatial ability and experience with instructional animations and static pictures were found to be significant predictors. The results suggest that gender and the type of spatial ability measures used both have a significant impact on gauging the effectiveness of instructional animations. Spatial ability measures should be tailored to gender and the specific nature of the learning domains to yield more consistent research results. (C) 2018 Elsevier Ltd. All rights reserved.