To establish whether plasma viscosity could be accurately measured from frozen-stored (instead of fresh) plasma samples in epidemiological studies, we prospectively compared fresh and frozen-stored samples from 1361 men and women aged 25-74 years participating in the Glasgow MONICA survey. Mean plasma viscosity (37 degrees C) was significantly higher (P < 0.0001) in fresh samples (mean = 1.310 mPa/s, standard deviation = 0.079 mPa/s) compared with frozen samples (mean = 1.278 mPa/s, standard deviation = 0.075 mPa/s). However, this difference (mean = 0.030 mPa/s, 95% confidence interval = 0.029-0.031) did not vary significantly with increasing viscosity (kappa statistic for consistency of categorization by thirds = 0.824, standard error = 0.021). We conclude that frozen samples are acceptable for epidemiological studies of plasma viscosity, which should increase their utilization.