Urinary phthalate excretion is used as marker of phthalate exposure in epidemiological studies. Here we examine the reliability of urinary phthalate levels in exposure classification by comparing the inter- and intrasubject variation of urinary phthalate metabolite levels. Thirty-three young healthy men each collected two spot, three first-morning, and three 24-h urine samples during a 3-month period. Samples were analyzed for the content of 12 urinary metabolites of 7 different phthalates. Variability was assessed as intraclass correlation coefficients (ICC). For the metabolites of diethyl-, dibutyl-, and butylbenzyl-phthalates moderate ICCs were observed in all three sample types, albeit highest in 24-h urine (0.51-0.59). For the metabolites of di(2-ethylhexyl) phthalate and di-iso-nonyl phthlates lower ICCs (0.06-0.29) were found. These low ICCs indicate a high risk of misclassification of exposures for these two phthalates in population studies and hence an attenuation of the power to detect possible exposure-outcome associations. The only slightly higher ICCs for 24-h pools compared to first-morning and spot urine samples does not seem to justify the extra effort needed to collect 24-h pools.