This study examined whether a reduction of exposure to tenofovir disoproxil as on-demand PrEP was associated with improved renal safety.  Investigators classified participants into people who: (a) reported daily PrEP use at ≥ 75% of visits (n = 499); (b) reported on-demand use at ≥ 75% of visits (n = 494); and (c) reported alternating between on-demand and daily use and did not meet criteria for other classification (n = 260).  Median baseline eGFR was 104 mL/min/1.73m² and 43% had baseline risk factors of kidney dysfunction, i.e., age > 40 years and/or baseline eGFR < 90 mL/min/1.73m².  After 22.3 months mean follow-up time, mean AUC of the ΔeGFR was -1.09 in the daily PrEP group, -0.69 in the switches group, and +0.18 in the on-demand group.  Compared to daily PrEP, AUC of the ΔeGFR was significantly higher in the on-demand group (this persisted after adjusting for baseline age and eGFR).  In participants with baseline risk factors of kidney dysfunction, adjusted mean AUC of ΔeGFR was significantly greater in the on-demand group compared to the daily group.  Multivariate analysis indicated that age > 40 years, baseline eGFR ≥ 90 mL/min/1.73m², daily PrEP and switching were associated with a higher risk of negative eGFR variation.  Five participants (2 in the daily group, 2 in the on-demand group, and 1 in the switches group) experienced a > 25% reduction in eGFR on two consecutive measurements (incidence rate: 2.5 per 1000 person-years).  Fanconi syndrome was not reported.  Two participants stopped PrEP because of eGFR decreases but were subsequently re-challenged without incident.