Determination of uranium isotopes in urine is critical for distinguishing between exposure to natural and anthropogenic sources of uranium. In biomonitoring, the extremely low levels of total uranium [U] present in urine and the complexity of the urine matrix present challenges to accurate isotope ratio determination. We have validated a method for measurement of 234U/238U, 235U/238U and 236U/238U in small-volume, random urine samples at occupational exposure levels, as well as for measurement of 235U/238U at biomonitoring levels. The method has been developed to detect exposure to mixtures of enriched and depleted uranium (DU) among former workers and nearby residents of the NL Industries factory in Colonie, NY, USA. Validation experiments employing solid phase extraction followed by SF-ICP-MS analysis were performed on base urine spiked with isotopic certified reference materials, isotopically depleted urine quality control materials, and natural uranium urine pools. Our results show that extraction with UTEVA resin yields lower blank levels than with TRU resin, and that double washing of the UTEVA resin columns results in further lowering of the blank. At occupational exposure levels ([U] = 50 ng L−1), 234U/238U, 235U/238U, and 236U/238U can be quantified in 1 mL of urine containing 3% DU with relative precisions (1s) of 10%, 0.4%, and 18%, respectively. For urine containing 75% DU and [U] = 50 ng L−1, the relative precision (1s) of 236U/238U is improved to <3%. At biomonitoring levels ([U] = 5 ng L−1), by increasing the urine volume to 4 mL, 234U/238U and 235U/238U can be quantified with relative precisions (1s) of 19% and 0.5%, respectively, whereas 236U/238U cannot be detected. By further increasing urine volume to 8 mL, comparable levels of isotope ratio precision are possible at [U] as low as 1 ng L−1. [ABSTRACT FROM AUTHOR]