Testosterone is involved in many processes like aggression and mood disorders. As it may easily diffuse from blood into saliva, salivary testosterone is thought to reflect plasma free testosterone level. If so, it would provide a welcome noninvasive and less stressful alternative to blood sampling. Past research did not reveal consensus regarding the strength of the association, but sample sizes were small. This study aimed to analyse the association in a large cohort. In total, 2,048 participants (age range 18–65 years; 696 males and 1,352 females) were included and saliva (using cotton Salivettes) and plasma were collected for testosterone measurements. Levels were determined by enzyme-linked immunosorbent assay and radioimmunoassay respectively. Free testosterone was calculated by the Vermeulen algorithm. Associations were determined using linear regression analyses. Plasma total and free testosterone showed a significant association with salivary testosterone in men (adjusted β = .09, p = .01; and β = .15, p < .001, respectively) and in women (adjusted β = .08, p = .004; and crude β = .09, p = .002 respectively). The modest associations indicate that there are many influencing factors of both technical and biological origin.