Objective. To describe the location and pattern of knee pain in patients with chronic, frequent knee pain using the Knee Pain Map, and to evaluate the inter- and intrarater reliability of the map. Methods. A cohort of 799 participants from the University of Pittsburgh Osteoarthritis Initiative Clinical Center who had knee pain in the last 12 months were studied. Trained interviewers assessed and recorded participant-reported knee pain patterns into 8 local areas, 4 regional areas, or as diffuse. Inter- and intrarater reliability were assessed using Fleiss' kappa. Results. Participants most often reported localized (69%) followed by regional (14%) or diffuse (10%) knee pain. In those with localized pain, the most commonly reported locations were the medial (56%) and lateral (43%) joint lines. In those with regional pain, the most commonly reported regions were the patella (44%) and medial region (38%). There was excellent interrater reliability for the identification of localized and regional pain patterns (κ = 0.7- 0.9 and 0.7- 0.8, respectively). The interrater reliability for specific locations was also excellent (κ = 0.7-1.0) when the number of participants with pain in a location was >4. For regional pain, the kappa for specific regions varied from 0.7-1.0. Conclusion. The majority of participants could identify the location of their knee pain, and trained interviewers could reliably record those locations. The variation in locations suggests that there are multiple sources of pain in knee OA. Additional studies are needed to determine whether specific knee pain patterns correlate with discrete pathologic findings on radiographs or magnetic resonance images.