Several scales are available for measuring degrees of liking/disliking. Information on characteristics of hedonic ratings as affected by scale types/lengths and overall product impression is limited. Therefore, we compared discriminatory power, sensitivity, reliablity, and neutral tendency of hedonic ratings collected from 9-point categorical (CAT), line (LIN), and labeled affective magnitude (LAM) scales (100-mm compared 300-mm length). Three grape juices (classified as A=liked, B=moderate, C=disliked) were arranged in 4 sets (AB-AC, BA-CA, AC-AB, or CA-BA; the left sample served first). Each panelist (N=60) evaluated color (OC), taste (OT), and overall-liking (OL) of 1 set (of 4 possible random serving sets) in all 6 independent sessions (3 scale types x 2 lengths). For OL determined on a 100-mm scale, the order of discriminatory power (determined by the MIXED procedure) was CAT>LAM>LIN. The 300-mm LAM scale was consistently more discriminatory than the 100-mm LAM scale for all attributes. For sensitivity (the number of significantly different pairs), CAT exhibited the highest sensitivity, considering simultaneously all products (A, B, C) categories. Regardless of attributes, product categories, and scale types, ratings tended to be lower with the 300-mm scale. For reliability (consistency of responses towards OC/OT/OL) of different scale/length types, Cronbach’s alpha values were 0.78-0.92, with slightly higher (0.89-0.92) values toward the disliked sample. Another reliability index (consistency of responses between 2 identical test samples) across different scale/length types was exhibited by Pearson’s correlation coefficients (r=0.32-0.75), with higher values toward the LAM scale; OC (a less complicated attribute) was less affected (higher r values). Product categories and scale types affected neutral tendency of responses, with moderate and disliked products being more affected when evaluated on LAM scales. This study demonstrated characteristics of hedonic scales as affected by scale lengths and overall product impression. Using proper scales to assess liking/disliking would help increase power of the experiment.