The Depression, Anxiety and Stress Scales are widely used in clinical and non-clinical populations, both in research and clinical settings. The need for briefer but valid and reliable instruments has motivated the reduction of the original 42-item scale to a short 21-version. On Portuguese samples, Pais-Ribeiro et al. (2004) found that the original 3-factor solution (stress, anxiety and depression) explained 50.35% of the variance and in an exploratory analysis; Xavier et al. (2015) presented a two factor solution and a modified three-factor solution with a sample of pregnant women, both with adequate factors’ reliability (<0.70) and explaining above 50% of the variance.

Based on the previous results of factor analysis with Portuguese samples, the present study aimed to perform confirmatory factor analyses (using Mplus software) to evaluate which dimensional structure best fitted the data.

The sample comprised 234 students (78.2% female), between 18–26 years old (M=20.55; SD=1.66). Eighty-five percent of the participants were on their first three years of college education. Participants filled the Portuguese version of the DASS-21.

Our results showed that the original 3-factor structure had the best model fit [χ²₍₁₈₆₎=475.465, P<0.05; RMSEA=0.082, 90% CI=0.073–0.091; CFI=0.918; TLI=0.908; SRMR=0.05]. Good reliability was found for all subscales (0.92 for stress, 0.87 for anxiety and 0.91 for depression subscale).

The DASS-21 is a reliable instrument that, with student populations, seems to have better performance when used with a 3-factor structure. Further research is needed to confirm this structure in Portuguese clinical samples.