Objective: Systematically analyze the methods used by clinical trials published in the last 10 years to collect preterm newborn saliva to measure cortisol. Method: Three databases PubMed, Scopus and Embase were searched in December 2016. Two researchers independently selected studies by title, and the abstracts of the selected studies were read to determine which ones met the inclusion criteria. All clinical trials of hospitalized preterm newborns that collected saliva to measure salivary cortisol were included. Results: Two types of methods were used for collecting saliva: aspiration or absorption, using absorbent materials. All nine aspiration-based studies and the 24 absorptionbased studies managed to measure salivary cortisol in at least some samples. Twenty-one studies reported losing some samples. The most common reasons for sample losses were insufficient saliva. Meta-regression (meta-analysis) and random effects model investigated possible relationships between percentage of sample losses and the following variables at a significance level of 5%: collection method, collection material, collection time and type of assay. Aspiration resulted in fewer sample losses than absorption (p<0.0001). However, only 33.3% (3 of 9) of aspiration-based studies reported the number of sample losses against 81.8% (18 of 22) of absorption-based studies. The most used assays to measure salivary cortisol were ELISA and radioimmunoassay. Conclusion: Salivary cortisol measurement in preterm newborns may be useful for assessing and comparing the level of stress generated or relieved by different stimuli. Saliva collection methods have not been standardized, preventing the reproduction of studies that use these methods.