Abstract Background Nosocomial infection of health-care workers during outbreaks of respiratory infections (eg, influenza A H1N1 ) is a great concern for public health policy makers. WHO-defined aerosol generating procedures (AGPs; eg, bronchoscopy, non-invasive ventilation) are thought to increase risk of aerosol transmission to health-care workers, but data to quantify risk accurately are insufficient. This study aims to provide some evidence base behind the requirement for respirators with FFP2 protection level to be worn in the vicinity of patients infected with influenza. Methods We measured the amount of pandemic H1N1 (2009) RNA in aerosols in the vicinity of H1N1-positive patients undergoing AGPs to help to quantify the potential risk of transmission to health-care workers. Sampling was done in the 2009–10 influenza seasons in five hospitals in England. Peak sampling times corresponded with peak levels of influenza activity (Nov, 2009–Jan, 2010; and Nov, 2010–Jan, 2011). Patient inclusion criteria were new changes in chest radiograph in the presence of one or more of: central temperature 38°C or higher (38·5°C in children) or 36°C or lower, white blood cell count less than 4000 or more than 12–000 cells per μL, positive microbiology or virology from respiratory secretions, and mucopurulent secretions from the respiratory tract. Aerosol sampling was done with the May 3 stage impinger (which allows fractionation of air samples into three aerodynamic size ranges: 0·86–4·0, 4·0–7·3, and >7·3 μm) at head height about 1 m away from the patient. Air samples were taken while WHO-defined AGPs were occurring, and baseline samples were taken when no AGP was being done. Samples were returned frozen to the laboratory, concentrated by centrifugation, and assayed by quantitative rt-PCR. A sample was deemed positive for aerosolised influenza particles if influenza RNA was detected in particles in the aerodynamic size range 0·86–7·3 μm. We used a univariable logistic regression model, adjusted for repeated measurements on the same individual, to examine the relation between propensity to obtain a positive H1N1 aerosol sample and undertaking of AGPs. Exact binomial confidence intervals were calculated for unadjusted sample proportions. Findings We obtained 198 air samples from 39 patients with influenza H1N1 (2009) during 99 sampling windows. Most participants were male (25 of 39 [64%]) and older than 50 years (28 of 39 [72%]). Only four (10%) participants were younger than 5 years. Mean time of sampling was 5·5 days (SD 8·56) since diagnosis. Viral RNA was detected in 21 of 198 (10·6%, 95% CI 6·7–15·8) air samples from ten of 39 (26%) patients. 46 air samples were obtained while WHO-defined AGPs were being done, of which nine (19·6%, 95% CI 9·4–33·9) contained viral RNA. Of 152 samples taken when no AGP was being done, 12 (7·9%, 4·2–13·4) contained viral RNA. Although AGPs did not significantly increase the probability of sampling an H1N1 (2009) positive aerosol, the point estimate and associated 95% CI from the regression model indicated that risk is likely to be heightened (odds ratio 4·31, 95% CI 0·83–22·5). Interpretation Our findings show that H1N1 RNA aerosols were often associated with infected patients within intensive-care units and were more frequently detected when a WHO-defined AGP was being done. Our results should be interpreted with caution because of the small sample size and limitations of the study design, which could not control for all potential sources of bias. Nevertheless on the basis of our results, we believe that infection control guidance for the use of FFP2 respirators in the vicinity of H1N1-positive patients should stand, until further data can be obtained. Funding Health Protection Agency.