The normalized gain, "g", has been an important tool for the characterization of conceptual improvement in physics courses since its use in Hake's extensive study on conceptual learning in introductory physics. The normalized gain is calculated from the score on a pre-test administered before instruction and a post-test administered after instruction and is defined as g = post-test - pre-test/100 - pre-test, (1) where both the pre-test and post-test have a maximum score of 100. The statistic has been used in many published works since Hake's paper. It has become sufficiently important that extensions to the statistic and investigations of its detailed properties have recently been published. This paper investigates the effect of students' guessing on the normalized gain and develops a correction for guessing for the pre-test and post-test. The normalized gain is found to be insensitive to the effects of guessing.