Summary: in situations where health workers and the public are exposed to pathogens, and commercial-grade masks are not available, it is possible to decrease the risk of infection using homemade masks using materials commonly available in the home. This brief review shows the level of filtering of pathogen-containing aerosols of homemade masks, and compares them to commercial surgical masks and N95-equivalent masks. The literature reviewed concludes that although surgical masks have a higher filtering or protection factor than homemade masks (by a factor of ~ 2), homemade masks are helpful in cases where no commercial units are available, and certainly better than wearing no protection at all. A sample design of a homemade mask is given in one of the articles and reproduced here for illustration. Links to the original papers are provided in the references section.
An article by van der Sande et. al. (Netherlands) assessed the transmission reduction provided by personal respirators, surgical masks and homemade masks (the latter made from teabag cloth.) They conclude that any type of general mask use is likely to decrease viral exposure and infection risk on a population, despite imperfect fits and imperfect adherence, with personal respirators (N95 masks) providing the most protection.
Personal respirators (N95) such as those worn by staff attending patients with respiratory infections are used primarily to protect the wearer; by contrast, surgical masks are primarily used to protect the environment from the respiratory droplets produced by the wearer. However, surgical masks are the ones typically available to the general population.
Masks efficacy is typically quantified by measuring the amount of particles concentrated on both sides of the mask, which are then analyzed using a particle counter (for particle size ranging between 0.02 - 1.00 microns, which encompasses most sizes of infectious respiratory aerosols.) The protection factor (also called "Fit Factor") PF is defined as the ratio of particle concentration on the outside of the mask to particle concentration on the inside of the mask. Thus, a high PF is always desirable. A PF of 1 means no protection at all.
In the above report, surgical masks provided about twice (average PF= 4.2) as much protection as homemade masks (average PF= 2.4); N95-equivalent masks provide about 40 times as much protection as homemade masks (average PF ~ 100).
Another simple homemade respirator mask made from heavy-weight T-shirt cotton (Honduras-made Hanes 100% pre-shrunk cotton t-shirt) was reported by V.M. Dato et. al. Prototypes were made with 8 layers of cloth attached to an additional patterned 37 cm x 72 cm cloth (see illustration below.) With this mask, the authors report PF's of 13 and 17, and up to 65 in one case (compared to PF ~ 100 for N95-equivalent commercial masks).
From V. M. Dato et. al., reference 3
In yet another study, cotton t-shirt masks were evaluated for their capacity to block bacterial and viral aerosols. The number of microorganisms isolated from coughs of healthy volunteers wearing their homemade mask, a surgical mask, or no mask was compared using several air-sampling techniques. The median protection factor for homemade masks was about 1/2 of surgical masks, consistent with reference .
Both masks significantly reduced the number of microorganisms emitted by volunteer individuals, although the surgical mask was 3 times more effective in blocking transmission than the homemade mask. The authors conclude that, while a homemade mask should only be considered as a last resort to prevent droplet transmission from infected individuals, it is better than no protection at all. They also found that 100% t-shirt cotton is the most suitable household material for improvised face masks because of its elasticity/pliability, providing a better fit to the face. The protection factor (PF) was similar to those from the Netherland's group (2.2 - 2.5 and 6 for homemade and surgical mask, respectively). They note that a close fit of the mask to the user's face, around mouth and nose, is as important as the filtering properties of the material, in order for it to be effective. Homemade masks can be tailored to each user's face structure and they can be washed and reused.
Conclusion: All reports coincide in that an improvised face mask should be viewed as a last resort, when a supply of commercial face masks is not available. While they are inferior to surgical masks and certainly personal filtering masks (N95), they can still help decrease transmission compared to wearing no mask at all.
 M. van der Sande et. al., PLoS ONE, 3(7), 2008. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2440799/
 TD Cerise Multi tea cloth, from Blokker.
 Virginia M. Dato et. al., Emerging Infectious Diseases, 12(6), 2006. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3373043/
 A. Davies, et. al. Disaster Medicine and Public Health Preparedness, August 2013.