The stability of a two-factor model recently proposed for the Gibb Experimental Test of Testwiseness was assessed, using confirmatory factor analysis. Designed to measure seven specific testwiseness skills with 10 items per skill, Gibb's test has been shown to discriminate between persons trained and untrained in selected testwiseness skills. Such a measure would have greater utility if the structure of the test were identified. Participants were 173 undergraduates. Confirmatory factor analyses were performed with LISREL 8 using total scores on the seven skills. Results indicated that the data fit the two-factor model and the simpler one-factor model. For this sample, the Gibb test could be characterized as tapping a general proficiency in testwiseness. Confirmation of the parsimonious one-factor model supports use of total scores from Gibb's test, although sampling fluctuation may be a concern. Gibb's test appears amenable to yielding a shorter form with fewer subscores or scales, which should facilitate measurement of testwiseness in future studies of training programs. Three tables present analysis results. (Contains 25 references.) (SLD)