Background: Recently, measurements of steroids like testosterone, androstenedione, cortisol and cortisone in saliva are more and more applied in diagnostics and scientific studies. This is mainly due to the simple and non-invasive collection of saliva. We aimed to evaluate the optimal way to collect saliva for steroid hormone measurement. Methods: We investigated in twenty volunteers whether there is a difference between steroid hormone concentrations in unstimulated and stimulated saliva collected while chewing, using cotton and synthetic Salivettes®, citric acid or chewing gum. Furthermore, total unstimulated saliva was compared to parotid gland saliva. Testosterone, androstenedione, cortisol and cortisone were measured using Liquid-Chromatography Tandem Mass Spectrometry (LC-MS/MS). Results: Salivary testosterone, androstenedione and cortisol concentrations were unaffected by stimulation upon mouth and tongue movements, cortisone levels were on average 16% lower. Concentrations of all hormones were lower in parotid gland saliva compared to total unstimulated saliva (on average 51%, 26%, 66% and 49% lower, for testosterone, androstenedione, cortisol and cortisone, respectively). Concentrations of testosterone as well as androstenedione were lower when using synthetic Salivettes® (58% and 41%, respectively) and were higher when using cotton Salivettes® (217% and 46%, respectively). Cortisol levels in saliva were unaffected by using Salivettes®. However, cortisol and testosterone levels were higher in with chewing gum stimulated saliva (16% and 55%, respectively). Cortisone concentrations were lower in all types of stimulations (on average 25%–35%). Conclusion: The way saliva is collected should be considered when analysing and interpreting salivary hormone concentrations. We advocate unstimulated saliva collection in simple polypropylene tubes for all steroid measurements.