Purpose of investigation: To compare whether albumin/creatinine ratios obtained from random or 8-hour urine collected in different periods of day differ in prediction of albumin excretion ≥2 g in 24-hour urine collection in preeclampsia. Methods: From a total of 70 women, 24-hour urine collected by three consecutive periods of eight hours and three random urine samples were taken before each period. The variation of albumin-creatinine ratios in samples across the day was analyzed by the Friedman and inter-assay coefficient variation. For each sample, receiver operator characteristic (ROC) curves were constructed to determine an optimal albumin/creatinine ratio value in the prediction of albuminuria ≥2 g. Results: The albumin/creatinine ratio did not vary significantly over time when all samples pooled. However, there was considerable intra-individual variation in both random and timed urine samples. On ROC analysis, the albumin/creatinine ratio in both random and timed urine samples predicted the 24-hour urine results and there was no difference between samples in prediction of albuminuria ≥2 g. A single optimal cut-off point was not available between samples. The positive and negative predictive values for optimal cut-offs ranged from 48%-88% and 94%- 100%, respectively. Conclusions: The random urine albumin/creatinine ratio was a poor predictor for proteinuria ≥2 g in patients with preeclampsia.