Differences between four test-independent components of testwiseness and their relative importance were studied in a quasi-experimental investigation of their effects. The components were: (1) time-using training; (2) error-avoidance training; (3) guessing training; and (4) deductive-reasoning training. Three parallel forms of a 30-item test were developed to measure the dependent variable--test performance. High school graduates (n=126) in a 6-week college preparatory program participated in the study. All participants (aged 15-20 years) received training in all components in different order, one each week. A control group received no training. Results indicate that although testwiseness does affect test performance, training in only one component is not sufficient to improve performance. There were no significant differences among the effects of the four components. Members of the training groups started outperforming the control group upon receiving training in at least two components. If only two components can be included in training, one skill should be related to not losing score points and one related to gaining extra points (e.g., error-avoidance and guessing). If three components are chosen, they should be the two skills related to gaining extra points (e.g., guessing and deductive reasoning), and one related to not losing score points (e.g., error-avoidance). A table shows the means and standard deviations of test performance scores. (SLD)