Introduction Over the last decades, the possibilities for monitoring the presence of drugs of abuse in oral fluid (OF) has increased enormously. Especially in the domain of testing drivers under influence of drugs (DUID), OF has gained popularity due to its non-invasive collected, directly on-site without hampering privacy. Several on-site immunosorbent tests (OIT) were developed over the years. For confirmatory analysis, the use of OF has been hampered by the influence of the salivary composition on the final drug concentrations. The OF collection protocol, the degree of stimulation of salivary flow, the physicochemical properties of drugs of abuse, and the type of drug administration are some of the parameters that influence the OF drug concentrations. Moreover, the final analytical result will be influenced by the OF collection. As a result, toxicological analysis of OF samples and the final interpretation is not always straightforward. Methods Laboratory and roadside studies, as well as double-blind placebo-controlled studies are used to evaluate the problems encountered using OF as confirmation matrix. The focus is put on Δ 9 -tetrahydrocannabinol (THC) analysis. The placebo-controlled study involves 10 test subjects smoking two subsequent doses of THC; 300 μg/kg and 150 μg/kg with a pause of 75 minutes using a Volcano vaporizer. Using the above mentioned studies, OIT (Drug-Test ® , Drager and Drugwipe ® , Securetec) and several OF collection devices (StatSure ® , Diagnostic Systems Inc; Quantisal ® , Immunalysis and Certus ® , Concateno) will be evaluated via UPLC-MS/MS analysis. Results Sensitivity of OIT seem to increase. The DrugTest 5000 and Drugwipe demonstrated respectively a sensitivity of 93% and 72% for Δ 9 -tetrahydrocannabinol during roadside studies. The Drugwipe is very sensitive just after smoking, but it rapidly decreases within 1.5 hours. The type of OF collection has to be specified when using OF as confirmation matrix, as THC recovery ranged from 50 to 70% and is dependent on the storage conditions. While matrix effects due to the collector buffers are observed, stability is often guaranteed via the stabilizing buffers. In a roadside study, we observed that after 5 minutes collection time, about 0.11–1.15 mL of OF was collected. Therefore, the final drug concentrations in neat OF have to be calculated via the following formula: C = [C UPLC/MS · (V1 + (W1–W2))] / [V2 x (W1–W2)] with C being the drug concentration in neat OF, C UPLC/MS the concentration obtained via the calibration curve, V1 the buffer volume, V2 the theoretical total collected volume and W1 and W2 respectively the weight of the collection device after and before OF collection. Neat OF THC concentrations in the placebo controlled study ranged from 12,361 ng/g 5 minutes after smoking down to 34 ng/g 80 minutes after 2 smoking sessions. Under placebo conditions, a median of 8 ng/g THC in OF was observed. The impact of all these observations on DUID legislations will be discussed. Conclusion OF has its limitations as a toxicological matrix. However, in situations where ‘recent’ drug use in combination with analytical cut-offs for interpretations are used, OF combines an easy collection of samples with a quick analysis of large batches.