The aim of the paper is to describe the process of development and evaluation of a newly designed spatial ability test. It consists of two consecutive studies. In Study I (N = 267) we proposed 35 items equally divided into five subsets. The items were designed with respect to theoretically described spatial ability dimensions (spatial perception, orientation, visualization, relations, and mental rotation). Even though the five factor structural model fitted the data reasonably well, on the principle of parsimony we agreed on un-dimensional model. Items with the best parameters (n = 25) were considered as the final version of the test. In Study II (N = 124) we verified that there is no significant impact of the administration media (paper/pencil vs. computer-based). The test-retest stability with a six-week interval was acceptable (r = 0.796), and so was the internal consistency (Cronbach’s α = 0.752). We have found a modest correlation (r = 0.470) with the Spatial Reasoning subtest of the Intelligence Structure Test.