Objectives To assess the association of fluoroquinolone use with tendon ruptures compared with no fluoroquinolone and that of the four most commonly prescribed non-fluoroquinolone antibiotics in the USA.

Design Retrospective observational study.

Setting US seniors enrolled in the federal old-age, survivor’s insurance programme.

Participants 1 009 925 Medicare fee-for-service beneficiaries and their inpatient, outpatient, prescription drug records were used.

Interventions Seven oral antibiotics, fluoroquinolones (ciprofloxacin, levofloxacin, moxifloxacin) and amoxicillin, amoxicillin-clavulanate, azithromycin and cephalexin.

Primary and secondary outcome measures All tendon ruptures combined, and three types of tendon ruptures by anatomic site, Achilles tendon rupture, rupture of rotator cuff and other tendon ruptures occurred in 2007–2016.

Results Of three fluoroquinolones, only levofloxacin exhibited a significant increased risk of tendon ruptures—16% (HR=1.16; 95% CI 1.06 to 1.28), and 120% (HR=2.20; 95% CI 1.50 to 3.24) for rotator cuff and Achilles tendon rupture, respectively, in the ≤30 days window. Ciprofloxacin (HR=0.96; 95% CI 0.89 to 1.03) and moxifloxacin (HR=0.59; 95% CI 0.37 to 0.93) exhibited no increased risk of tendon ruptures combined.

Among the non-fluoroquinolone antibiotics, cephalexin exhibited increased risk of combined tendon ruptures (HR=1.31; 95% CI 1.22 to 1.41) and modest to large risks across all anatomic rupture sites (HRs 1.19–1.93) at ≤30 days window. Notably, the risk of levofloxacin never exceeded the risk of the non-fluoroquinolone, cephalexin in any comparison.

Conclusions In our study, fluoroquinolones as a class were not associated with the increased risk of tendon ruptures. Neither ciprofloxacin nor moxifloxacin exhibited any risk for tendon ruptures. Levofloxacin did exhibit significant increased risk. Cephalexin with no reported effect on metalloprotease activity had an equal or greater risk than levofloxacin; so we question whether metalloprotease activity has any relevance to observed associations with tendon rupture. Confounding by indication bias may be more relevant and should be given more consideration as explanation for significant associations in observational studies of tendon rupture.