Two Cobas-Bio analyzers were found to have significantly increased imprecision, owing to absorbance errors, when we used rotors pieced together from segments of four to 10 cuvets. The difference in CV between whole/unbroken and four-cuvet or 10-cuvet segment rotors was significant by Student's unpaired t-test (P = 0.038 and less than 0.0001, respectively) and by the Mann-Whitney U test (P less than 0.0001 for either). Pieced-together rotors were found to have plastic fragments in some cuvets, both adjacent to the break points and in other positions within the segment. There was no significant difference in delta A between adjacent cuvets at break positions or in between break positions by unpaired t-test (P = 0.214) or by the Mann-Whitney U test (P = 0.235). There was a significant difference by either test (P less than 0.0001) for the delta A between adjacent cuvets in broken rotors and those in whole/unbroken rotors. We conclude that random errors in absorbance are observed both in cuvets adjacent to break positions and in cuvets removed from the break point when pieced-together rotors are used.