Concerns about the likely occurrence of an influenza pandemic in the near future are increasing. The highly pathogenic strains of influenza A (H5N1) virus circulating in Asia, Europe, and Africa have become the most feared candidates for giving rise to a pandemic strain. Several authors have stated that large-droplet transmission is the predominant mode by which influenza virus infection is acquired ( 1 – 3 ). As a consequence of this opinion, protection against infectious aerosols is often ignored for influenza, including in the context of influenza pandemic preparedness. For example, the Canadian Pandemic Influenza Plan and the US Department of Health and Human Services Pandemic Influenza Plan ( 4 , 5 ) recommend surgical masks, not N95 respirators, as part of personal protective equipment (PPE) for routine patient care. This position contradicts the knowledge on influenza virus transmission accumulated in the past several decades. Indeed, the relevant chapters of many reference books, written by recognized authorities, refer to aerosols as an important mode of transmission for influenza ( 6 – 9 ). In preparation for a possible pandemic caused by a highly lethal virus such as influenza A (H5N1), making the assumption that the role of aerosols in transmission of this virus will be similar to their role in the transmission of known human influenza viruses would seem rational. Because infection with influenza A (H5N1) virus is associated with high death rates and because healthcare workers cannot as yet be protected by vaccination, recommending an enhanced level of protection, including the use of N95 respirators as part of PPE, is important. Following are a brief review of the relevant published findings that support the importance of aerosol transmission of influenza and a brief discussion on the implications of these findings on pandemic preparedness. Influenza Virus Aerosols By definition, aerosols are suspensions in air (or in a gas) of solid or liquid particles, small enough that they remain airborne for prolonged periods because of their low settling velocity. For spherical particles of unit density, settling times (for a 3-m fall) for specific diameters are 10 s for 100 μm, 4 min for 20 μm, 17 min for 10 μm, and 62 min for 5 μm; particles with a diameter 6-μm diameter are trapped increasingly in the upper respiratory tract ( 12 ); no substantial deposition in the lower respiratory tract occurs at >20 μm ( 11 , 12 ). Many authors adopt a size cutoff of 10–20 μm will settle rapidly, will not be deposited in the lower respiratory tract, and are referred to as large droplets ( 10 – 12 ). Coughing or sneezing generates a substantial quantity of particles, a large number of which are 40%. The increased survival of influenza virus in aerosols at low relative humidity has been suggested as a factor that accounts for the seasonality of influenza ( 15 , 16 ). The sharply increased decay of infectivity at high humidity has also been observed for other enveloped viruses (e.g., measles virus); in contrast, exactly the opposite relationship has been shown for some nonenveloped viruses (e.g., poliovirus) ( 11 , 15 , 16 ). Experimental Influenza Infection Experimental infection studies permit the clear separation of the aerosol route of transmission from transmission by large droplets. Laboratory preparation of homogeneous small particle aerosols free of large droplets is readily achieved ( 13 , 18 ). Conversely, transmission by large droplets without accompanying aerosols can be achieved by intranasal drop inoculation ( 13 ). Influenza infection has been documented by aerosol exposure in the mouse model, the squirrel monkey model, and human volunteers ( 12 , 13 , 17 – 19 ). Observations made during experimental infections with human volunteers are particularly interesting and relevant. In studies conducted by Alford and colleagues ( 18 ), volunteers were exposed to carefully titrated aerosolized influenza virus suspensions by inhaling 10 L of aerosol through a face mask. The diameter of the aerosol particles was 1 μm–3 μm. Demonstration of infection in participants in the study was achieved by recovery of infectious viruses from throat swabs, taken daily, or by seroconversion, i.e., development of neutralizing antibodies. The use of carefully titrated viral stocks enabled the determination of the minimal infectious dose by aerosol inoculation. For volunteers who lacked detectable neutralizing antibodies at the onset, the 50% human infectious dose (HID50) was 0.6–3.0 TCID50, if one assumes a retention of 60% of the inhaled particles (18). In contrast, the HID50 measured when inoculation was performed by intranasal drops was 127–320 TCID50 ( 13 ). Additional data from experiments conducted with aerosolized influenza virus (average diameter 1.5 μm) showed that when a dose of 3 TCID50 was inhaled, ≈1 TCID50 only was deposited in the nose ( 12 ). Since the dose deposited in the nose is largely below the minimal dose required by intranasal inoculation, this would indicate that the preferred site of infection initiation during aerosol inoculation is the lower respiratory tract. Another relevant observation is that whereas the clinical symptoms initiated by aerosol inoculation covered the spectrum of symptoms seen in natural infections, the disease observed in study participants infected experimentally by intranasal drops was milder, with a longer incubation time and usually no involvement of the lower respiratory tract ( 13 , 20 ). For safety reasons, this finding led to the adoption of intranasal drop inoculation as the standard procedure in human experimental infections with influenza virus ( 13 ). Additional support for the view that the lower respiratory tract (which is most efficiently reached by the aerosol route) is the preferred site of infection is provided by studies on the use of zanamivir for prophylaxis. In experimental settings, intranasal zanamivir was protective against experimental inoculation with influenza virus in intranasal drops ( 21 ). However, in studies on prophylaxis of natural infection, intranasally applied zanamivir was not protective ( 22 ), whereas inhaled zanamivir was protective in one study ( 23 ) and a protective effect approached statistical significance in another study ( 22 ). These experiments and observations strongly support the view that many, possibly most, natural influenza infections occur by the aerosol route and that the lower respiratory tract may be the preferred site of initiation of the infection. Epidemiologic Observations In natural infections, the postulated modes of transmission have included aerosols, large droplets, and direct contact with secretions or fomites because the virus can remain infectious on nonporous dry surfaces for >(January 2006) recommends FFP2 respirators (equivalent to N95 respirators) (http://www.splf.org/s/IMG/pdf/plan-grip-janvier06.pdf). Given the scientific evidence that supports the occurrence of aerosol transmission of influenza, carefully reexamining current recommendations for PPE equipment would appear necessary.