The identification and recall of subsea markers is a common task amongst offshore commercial divers. Two experiments explored the accuracy of reproduction of subsea marker panels by commercial divers and correlations with tests of embedded figures, memory-for-designs, spatial relations, dotted outlines, and block design. Trials were conducted in diving tanks and in open water. In the diving tanks 30% recall errors occurred and performance correlated r= 0·75 with the embedded figures test. In the open sea again 30% recall errors were found and performance was well predicted by the embedded figures test (r= 0·69). All other tests failed to predict performance when the effects of the embedded figures were removed.