Several playing experiments have been reported in recent years to assess violin qualities but the results have shown large inter-subject inconsistencies. A new perceptual playing experiment was conducted to examine how violinists might differentiate between "good" and "bad" violins, and to what extent they might agree with each other. A pool of six violins of was assembled: three performance violins and three student violins. Nine violinists participated: 3 professional violinists, 1 graduate student, 4 undergraduate students in music performance and 1 had a conservatory degree. The experiment was organized in two phases. During the first phase, subjects were asked to rate and rank the six violins from least preferred to most preferred on a continuous scale. During the second phase, subjects were asked to rate and rank the violins according to five criteria: responsiveness, resonance, clarity, richness, and balance. The results showed that the three performance violins were on average rated significantly higher than the three student violins in terms of preference and all five criteria except responsiveness. We also found that the violinists preferred violins with richer and clearer sound. Three professional violinists rated performance violins much higher than student violins.