Background and purpose — We aimed to examine the temporal trends in periprosthetic joint infection (PJI) revision incidence after knee arthroplasty (KA) from 1997 through 2019. Patients and methods — 115,120 primary KA cases from the Danish Knee Arthroplasty Register were followed until the first PJI revision. We computed cumulative inci-dences and adjusted hazard ratios (aHRs) of PJI revision by calendar periods and several patient-and surgical-related risk factors. Results were analyzed from 0–3 months and from 3–12 months after KA. Results — The overall 1-year PJI revision incidence was 0.7%, increasing from 0.5% to 0.7% (1997 through 2019). The incidence of PJI revision within 3 months increased from 0.1% to 0.5% (1997 through 2019). The adjusted hazard ratio (aHR) within 1 year of primary KA was 5.1 compar-ing 2017–2019 with 2001–2004. The PJI revision incidence from 3–12 months of KA decreased from 0.4% to 0.2%, with an aHR of 0.5 for 2017–2019 vs. 2001–2004. Male sex, age 75–84 (vs. 65–74), and extreme obesity (vs. normal weight) were positively associated with the risk of PJI revision within 3 months, whereas only male sex was associated from 3–12 months. Partial knee arthroplasty (PKA) vs. total KA was associated with a lower risk of PJI revision both within 3 months and 3–12 months of KA. Conclusion — We observed an increase in PJI revision within 3 months of KA, and a decrease in PJI revision incidence from 3–12 months from 1997 through 2019. The reasons for this observed time-trend are thought to be multifac-torial. PKA was associated with a lower risk of PJI revision.