In order to understand the reliability of tests of fearfulness, Japanese quail were tested and retested in four different tests: an emergence test, a novel object test, a novel food test, and a predator surprise test. The results indicated that some of the measures were not reliable. Some examples of inconsistencies included: 1) there was a low correlation between the two tests for latency to extend head into emergence arena during the emergence test, but the number of times they extended there heads into the arena was more consistent; and 2) latency to try a novel food had inconsistent results, but the amount of novel food eaten was more reliable. The use of partial correlations was found to be more useful when attempting to determine test-retest reliability, than Pearson’s coefficient. It appears that the common behavioural indicators of fearfulness do not always have consistent results.