Unexpected deviations of ex-core neutron detector signals were observed during a voyage of the Japanese nuclear ship, Mutsu. From detailed three-dimensional analyses, this phenomenon was determined to be caused by an asymmetrical neutron source distribution in the core due to a small misalignment between the two control rods of a control rod group. A systematic ex-core detector response experiment was performed during the Mutsu’s third experimental voyage to gain some understanding of the relationship between the control rod pattern and the detector response characteristics. Results obtained from analyses of the experiment indicate that the Crump-Lee technique, using calculated three-dimensional source distributions for various control rod patterns, provides good agreement between the calculated and measured detector responses. Xenon transient analyses were carried out to generate accurate three-dimensional source distributions for predicting the time-dependent detector response characteristics. Two types of ex-core detector responses are caused by changes in the control rod pattern in the Mutsu reactor: The detector response ratio tends to decrease with the withdrawal of a group of control rods as a pair, and a difference in the positions of the control rods in a group causes signal deviations among the four ex-core detectors. Control rod misalignment does not greatly affect the mean value of the four detector signals, and the deviation can be minimized if the two rods within a group are set at the same elevation at the time of detector calibration.