The environment has been shown to be a source of pathogens causing infections in hospitalised patients. Incorporation of pathogens into biofilms, contaminating dry hospital surfaces, prolongs their survival and renders them tolerant to normal hospital cleaning and disinfection procedures. Currently there is no standard method for testing efficacy of detergents and disinfectants against biofilm formed on dry surfaces. Aim: The aim of this study was to develop a reproducible method of producing Staphylococcus aureus biofilm with properties similar to those of biofilm obtained from dry hospital clinical surfaces, for use in efficacy testing of decontamination products. The properties (composition, architecture) of model biofilm and biofilm obtained from clinical dry surfaces within an intensive care unit were compared. Methods: The CDC Biofilm Reactor was adapted to create a dry surface biofilm model. S. aureus ATCC 25923 was grown on polycarbonate coupons. Alternating cycles of dehydration and hydration in tryptone soy broth (TSB) were performed over 12. days. Number of biofilm bacteria attached to individual coupons was determined by plate culture and the coefficient of variation (CV%) calculated. The DNA, glycoconjugates and protein content of the biofilm were determined by analysing biofilm stained with SYTO 60, Alexa-488-labelled Aleuria aurantia lectin and SyproOrange respectively using Image J and Imaris software. Biofilm architecture was analysed using live/dead staining and confocal microscopy (CM) and scanning electron microscopy (SEM). Model biofilm was compared to naturally formed biofilm containing S. aureus on dry clinical surfaces. Results: The CDC Biofilm reactor reproducibly formed a multi-layered, biofilm containing about 107CFU/coupon embedded in thick extracellular polymeric substances. Within run CV was 9.5% and the between run CV was 10.1%. Protein was the principal component of both the in vitro model biofilm and the biofilms found on clinical surfaces. Continued dehydration and ageing of the model biofilm for 30days increased the % of protein, marginally decreased gylcoconjugate % but reduced extracellular DNA by 2/3. The surface of both model and clinical biofilms was rough reflecting the heterogeneous nature of biofilm formation. The average maximum thickness was 30.74±2.1μm for the in vitro biofilm model and between 24 and 47μm for the clinical biofilms examined. Conclusion: The laboratory developed biofilm was similar to clinical biofilms in architecture and composition. We propose that this method is suitable for evaluating the efficacy of surface cleaners and disinfectants in removing biofilm formed on dry clinical surfaces as both within run and between run variation was low, and the required equipment is easy to use, cheap and readily available.