A total of 78 10th, 11th, and 12th grade smokers (at least one cigarette per week) who had completed both computer sessions were included in the analysis. Individuals who changed the number of cigarettes smoked between session I and II were considered to have shifted their "true score" on nicotine dependence. No proxy item for nicotine dependence was available in the survey, so the item "How many cigarettes do you smoke a day?" was extracted from the modified FTQ to serve as the proxy item; thus, the analysis of test-retest reliability was based on six FTQ items. The test- retest analysis was restricted to students whose number of cigarettes per day did not vary across time points (n=51). Principal components analysis on the seven-item FTQ resulted in a single dimension, accounting for 49.6 percent of the total variance. The six-item FTQ also appears to be unidimensional, explaining 52.5 percent of the total variance. The computed component loadings support equal rather than differential item weighting. Corrected item-total correlations for the seven-item scale ranged from 0.47 to 0.73; for the six-item FTQ, the corrected item-total correlations ranged from 0.45 to 0.71. Although the modified six-item FTQ revealed sufficient stability over a 2-month interval, with a test-retest correlation of 0.71, future research should include the development of items that better reflect adolescent smoking patterns. Limitations of the study are discussed. 1 figure, 3 tables, and 22 references