A comparison was made to determine whether thumb-counting or comb-counting was more accurate for determining flea infestation levels on dogs when performed for equal periods of time. To accomplish this, ten beagle dogs were each infested with 100 adult fleas, Ctenocephalides felis. After the fleas were allowed to disperse for 1 h the dogs were examined using the thumb-counting method. The time required to cover each dog and the number of fleas counted were recorded. Thumb-counting times ranged from 3.0 to 4.8 min. Each of the dogs was then examined by the comb-counting method for the same amount of time it had been thumb-counted. The thumb-counting method detected a geometric mean of five (range, 0-13) fleas per dog, while comb-counting recovered a mean of 73.5 (range, 57-87) fleas per dog. These results were significantly different (P < 0.01), indicating that the differences in accuracy previously recorded for the two methods are independent of time. The standard deviations for both methods were also statistically significantly different, suggesting that comb-counting is also more precise than the thumb-counting method.