We previously showed that chronic cocaine use by active illicit users produced a longer plasma half-life than expected based on acute low-dose cocaine studies. Here we report urinary excretion patterns of cocaine metabolites as benzoylecgonine (BE) equivalents from 18 of the same individuals, housed for up to 14 days on a closed research unit. In addition, we evaluated whether creatinine normalization of BE equivalents increased mean detection time and reduced mean within-subject variability. All urine voids (N = 953) were individually assayed; BE equivalents were determined semi-quantitatively by FPIA. Compared to concentration in first void after admission, BE equivalents decreased to approximately 33%, 8%, and 4% at 24, 48, and 72 h, respectively. Mean +/- SD (range) time to first negative specimen (BE equivalents < 300 ng/mL) was 43.6 +/- 17.1 (16-66) h. BE equivalents fluctuated considerably across successive specimens; 69% of participants tested positive at least once after testing negative, and the mean time to last positive specimen was 57.5 +/- 31.6 (11-147) h after the first specimen. Thus, mean cocaine metabolite detection times were consistent with prolonged elimination, with 63% of participants testing positive longer than the expected 48-h window of detection after admission to the unit. Mean time to last positive after last use of cocaine, known by self-report only, was approximately 81 +/- 34 (34-162) h. Creatinine normalization, with the cut-off of 300 ng BE equivalents/mg creatinine, increased detection time: mean time to first negative specimen was 54.8 +/- 20.7 (20-100) h, and mean time to last positive specimen was 88.4 +/- 51.0 (35.6-235) h. Compared with the concentration in the first void after admission, BE equivalents/creatinine decreased to approximately 56%, 6%, and 5% at 24, 48, and 72 h. However, creatinine normalization did not reduce the fluctuation of BE equivalents across successive specimens. Thus, creatinine normalized values may be useful when the goal is to maximize the probability or duration of cocaine metabolite detection, but may be less useful in determining whether an individual has used cocaine since a previous specimen collection.