In an in vitro demineralization model, the protective effect of two chlorhexidine varnishes, Cervitec (1% w/w chlorhexidine diacetate and 1% w/w thymol) and EC40 (40% w/w chlorhexidine diacetate), was compared with that of Fluor Protector, a varnish containing 0.1% w/w F. The demineralization model comprised an acidogenic Streptococcus mutans suspension in agarose placed on enamel or dentine specimens. The experiments extended over three serial 22-hour demineralization periods with fresh S. mutans suspensions for each period. To determine whether the varnishes released enough demineralization-inhibiting compounds, approximately 10 microliters of the varnishes was applied adjacent to the enamel and dentine specimens just before the first application of the S. mutans suspensions and left during the serial experiments (release study). In a separate series of experiments, the effect of the pretreatment of the enamal and dentine specimens with the various varnishes was tested (pretreatment study). In the release study the protective effect for enamel decreased in the order: EC40 = Fluor Protector >> Cervitec = no treatment. For dentine this order was: EC40 >> Fluor Protector = Cervitec > no treatment. In the pretreatment study, the enamel specimens were best protected by Fluor Protector (Fluor Protector >> Cervitec = EC40 > no treatment), while the dentine specimens were best protected by the chlorhexidine treatments (Cervitec = EC40 > Fluor Protector > no treatment). A 1:1 mixture of Cervitec and Fluor Protector was as effective as the most effective component alone. It is concluded that a varnish containing both fluoride and chlorhexidine may be useful, since it could give optimal protection to both enamel and dentine.