Graphical abstract. Aqueous Cotoneaster nummularia leaf extract was tested by broth microdilution against E. coli and P. aeruginosa from hospital waste; partial inhibition (47–72 %) was observed without reaching the MIC criterion. Escherichia coli and Pseudomonas aeruginosa are leading causes of drug-resistant infection and are listed among the World Health Organization priority pathogens. Species of Cotoneaster (Rosaceae) are used in traditional Asian medicine and are rich in polyphenols, but evidence for activity of Cotoneaster nummularia against Gram-negative bacteria of clinical or environmental origin is scarce. Leaves of C. nummularia were collected in Dir Lower, Khyber Pakhtunkhwa, Pakistan, and extracted with water (1g: 400 mL; 24 h maceration followed by 10 min at 80 °C). E. coli and P. aeruginosa were isolated from dry waste of Saidu Teaching Hospital, Swat. Growth inhibition was measured by 96-well broth microdilution over a two-fold series of 4.88–156.25 µg/mL (leaf-equivalent) by optical density at 600 nm after 24 h at 37 °C. The minimum inhibitory concentration (MIC) was defined a priori as the lowest concentration producing ≥ 90 % inhibition. Concentration–response association was assessed by Spearman correlation and between-isolate differences by the Wilcoxon signed-rank test. The extract partially inhibited both isolates at every concentration tested (E. coli 47.2–67.7 %; P. aeruginosa 52.6–71.8 %). No concentration reached 90 % inhibition; the MIC therefore exceeded 156.25 µg/mL for both organisms. Inhibition of ≥ 50 % was observed from 9.77 µg/mL for E. coli and at all concentrations for P. aeruginosa. Inhibition was not significantly associated with concentration (E. coli ρ = 0.77, p = 0.07; P. aeruginosa ρ = 0.26, p = 0.62), and mean inhibition did not differ between isolates (difference −1.5 percentage points; p = 0.84). Aqueous C. nummularia leaf extract exerts a partial, plateau-type growth-inhibitory effect on Gram-negative hospital-waste isolates but does not achieve MIC-level inhibition within the tested range.