Background: PSA levels are often verified before recommending a prostate biopsy because of the variability of PSA, We used data from the STHLM3 trial to assess the value of a second PSA test taken prior to biopsy. Methods: STHLM3 is a prospective and population-based prostate cancer diagnostic study performed in Stockholm, Sweden, 2013-2015. 1686 men in STHLM3 with a PSA value 3-10 ng/ml performed a second PSA test within 8 weeks of the first test and then underwent prostate biopsy. We compared proportions of cancer detected in men with stable PSA levels (2^nd PSA within +/-10% of 1^stPSA) to men whose PSA levels increased or decreased at least 10%, respectively. Results: Men with GS ≥ 7 cancer in the biopsy had more stable PSA than men with benign biopsies; the median change was 13.7% (inter-quartile range (IQR) 6.1-26.9) in men with benign biopsies, 13.0% (IQR 6.1-25.6) in men with GS ≤ 6 tumors and 8.2% (IQR 4.5-16.7) in men with GS ≥ 7 tumors. The average risk of having GS ≥ 7 cancer was 15.4% (95% CI 13.7 – 17.1) after the first test (i.e. the entire cohort), 21.6% (95% CI 18.6 – 24.6) in men with stable PSA levels compared with 13.5% (95% CI 10.2 – 16.7) and 9.0% (95% CI 6.6 – 11.4) if PSA levels increased or decreased at least 10%, respectively. There were no significant differences in risk of GS ≤ 6 tumor for men with stable PSA compared to men with increasing/decreasing PSA. Conclusions: As previously shown, the risk of prostate cancer decreases if a repeat PSA test shows decreasing levels. We show that this effect mainly affect GS ≥ 7 tumors, and that the risk of prostate cancer also decreases with increasing PSA values. Our findings need validation, but suggest that PSA is more stable in men with GS ≥ 7 cancer, whereas elevated but fluctuating PSA levels to a higher degree are caused by other processes (e.g. inflammation). Use of decision tools could aid incorporation of complex information such as repeat PSA values in diagnostics.