The purpose of the study was to investigate the problem of habit strength as indirectly measured by the Thorndike-Lorge frequency ratings. Four groups of 15 words, each group representing a different frequency category, were read to 56 college students. A second test was conducted with 16 graduate students in which 4 groups of 10 words each were read. The results of the first test showed that in recall there is a clear decline in the number of words recalled as a function of word frequency, with the exception of the difference between the number of words recalled in the last 2 frequency groups. In the first minute of recall the differences between the mean number of AA words (frequency greater than 100 per million) recalled and the number recalled from lower groups are all significant between the 0.05 and 0.01 levels. The superiority of high frequency recall was only maintained for the first minute of an 8-min test. The second test failed to substantiate the results of the first test, but none of the relationships were significant.