Background <p>Portable air cleaners (PACs) can be used to reduce indoor concentrations of fine particulate matter (PM_2.5), a harmful air pollutant associated with adverse health outcomes.</p> Objective <p>Evaluate the impact of environmental conditions and device operation on PAC performance and user perceptions of PACs and indoor air quality.</p> Methods <p>We evaluated PACs in 60 apartments in three multifamily buildings in Toronto, Canada, using a three-arm randomized crossover design. Participants experienced three experimental conditions, each for a period of one week: placebo (no air cleaning), constant (PAC always operating), and auto (PAC operating when concentrations exceeded 10 μg·m⁻³). PM_2.5 concentrations were continuously measured and participants completed surveys to report their behaviours and perceptions of the PAC during the previous week.</p> Results <p>During the placebo arm, weekly median concentrations ranged from 1 to 501 μg·m⁻³. During the constant arm, weekly median concentrations were reduced in all 60 homes by 1 to 285 μg·m⁻³ (absolute reduction) and 14% to 100% (relative reduction, or effectiveness). During the auto arm, weekly median concentrations were reduced in 42 of 59 homes, accompanied by substantial reductions in PAC runtimes relative to constant operation. Weekly mean concentrations were reduced by an average of 66% (95% CI: 59%–72%) during the constant arm and 40% (95% CI: 27%–50%) during the auto arm after adjusting for measured behaviours and unmeasured between-home differences using linear mixed-effects regression. Participant behaviours contributed to the large variation in concentrations between and within homes, which affected absolute reduction and effectiveness. A comparison of measured and theoretical performance suggests a discrepancy between room volume and the effective mixing volume of the environment being cleaned.</p> Significance <p>These results demonstrate that PAC performance can vary between homes because of environmental and behavioural factors that must be accounted for when interpreting the results of an intervention.</p> Impact <p>This study evaluates PAC performance across homes within multifamily buildings, allowing for a demonstration of how random variation between and within homes can affect measurements of PAC performance. The discrepancy between measured and theoretical performance has implications for PAC sizing guidelines. PAC automation can reduce device runtimes, addressing user concerns about noise, but a universal control strategy may not be suitable. With constant operation, PACs reduced weekly median concentrations in all. PACs may be particularly beneficial in multifamily buildings, where, compared with detached homes, PM_2.5 concentrations are often disproportionately high and where centralized air cleaning is less common.</p>