Background: The real-world impact of metastatic (met) site on Tx choice in mCRC is unknown. Therapeutic associations across metastases can identify gaps in care and inform clinical decision support systems. This study assessed Tx patterns in mCRC pts by met site. Methods: 4656 stage III/IV pts =18 yrs at mCRC diagnosis receiving anticancer agents were followed in a retrospective observational study using electronic medical records from the largest oncology database (SDI) from 1/1/04-7/31/11. 1st- (1L), 2nd- (2L), and 3rd-line (3L) therapies were identified for top 4 met sites. Logistic regression adjusting for met site overlap determined Tx use by met site and therapy line. Results: Pts were mostly male (54.1%) and on average 61.6 yrs. The top 4 met sites were liver (34.2%), lung (15.9%), lymph (15.8%), and bone (7.5%). Leading regimens are presented in the table below. Bevacizumab (B) use varied by met site and therapy line. Against other/unknown met sites, 1L B was more common in liver (OR=2.06 [1.75-2.42]) and bone (OR=1.46 [1.16-1.84]), less common in lymph (OR=0.60 [0.50-0.73]), and statistically equal in lung (OR=1.01 [0.85-1.20]). In 2L, B was more likely to be given for liver (OR=1.77 [1.48-2.11]) and lung (OR=1.45 [1.21-1.74]) vs no significant difference for bone (OR=1.17 [0.92-1.50]) and lymph (OR=0.87 [0.71-1.07]) sites. B use in 3L mirrored 2L. Conclusions: Besides 1L in pts with lymph metastases, top regimens across met sites for 1L, 2L, and 3L were FX + B, FR + B, and IRI + C, respectively; however, significant differences in B Tx patterns were evident. Further research should examine the impact of differential B use on outcomes.