PURPOSE: To assess the reproducibility of one-, two-, and three-dimensional measurements of the size of liver metastases. MATERIALS AND METHODS: The sizes of 10 liver metastases were determined 200 times: Ten observers measured each lesion by using images obtained at two different CT studies and with five different measurement techniques (maximum diameter, product of diameters, area, volume, and product of three diameters). They repeated each measurement in a separate session. The influence of measurement technique and lesion type (size and morphologic appearance) on measurement reproducibility was assessed. RESULTS: Three-dimensional measurements proved to be as reproducible as one- and two-dimensional measurements. Measurement reproducibility was influenced mainly by the size (P < .0001) and morphologic appearance (P < .01) of the lesions. CONCLUSION: Three-dimensional size measurements obtained with spiral CT are reproducible and could therefore replace the two-dimensional measurements of tumor size currently obtained with follow-up CT studies.