Comb-counting and thumb-counting were compared in a cross-over study to determine which was more accurate for quantifying flea infestation levels on dogs. Twenty beagle dogs were used in the study and infested with either 50 or 100 adult fleas (Ctenocephalides felis). Two groups of five dogs each were infested with either 50 or 100 fleas per dog, and then comb-counted with a fine-toothed flea comb for 8 min periods. An additional two groups of five dogs each were also given 50 or 100 fleas, and then thumb-counted. The counting time for this technique is both lower and more variable because the fleas are only observed and not captured; thus, the speed at which the dog is covered must be increased in order to prevent counting the same fleas more than once. The mean time of thumb-counting per dog was 3.2 min. Fleas removed during comb-counting were placed back on the dog they were taken from after the count was concluded. At the cross-over point, the ten dogs that had been comb-counted were then thumb-counted and the ten dogs that had been thumb-counted were comb-counted. The results showed that comb-counting recovered significantly (P < or = 0.05) more fleas than did thumb-counting. On dogs given 50 and 100 fleas, comb-counting gave mean percentage recoveries of 67.6% and 75.4%, respectively, whereas thumb-counting found means of 8.8% and 7.7%, respectively. The order in which the counting methods were employed produced no significant effect (P > 0.05) on the number of fleas counted.