The construct validity of three variants of the Trail Making Test was investigated using 162 undergraduate psychology students. During a 3-week period, the Trail Making Test of the Delis-Kaplan Executive Function System, Comprehensive Trail Making Test, and Connections Task were administered in six possible orders. Using confirmatory factor analysis, a two-factor (sequencing, shifting) model was identified as the best fitting model for the data as an alternative to unitary and three-factor (sequencing, shifting, scanning) models. The two-factor structure was invariant across groups. A latent means structural analysis yielded no differences between the factor means for each of the groups, which indicates the absence of order effects and supports the interchangeable use of the three tests. The study proposes the use of alternate, equivalent tasks to help eliminate the effects of practice associated with serial assessment and should be replicated in a clinical sample. (Contains 5 tables, 1 figure, and 1 note.)