At the end of 2019, the first reports of an outbreak of a new coronavirus called SARS-CoV-2 appeared in Wuhan province, China. The virus has spread rapidly around the world, and it became one of the greatest global health crises of the 21st century. Mass population testing is key to preventing the spread of the covid-19 pandemic. RT-qPCR, the golden standard of molecular testing, has limitations that have become apparent mainly due to the significant increase in need for testing. RT-qPCR is time consuming, requires a well equipped diagnostic laboratory, and in the case of mass testing, in which we all rely on the same method, can lead to a shortage in commercial reagents and equipment. Due to ease of use, speed, cost-effectiveness, and because the reaction does not require a rapid change in temperature for amplification, reverse transcription and loop-mediated isothermal amplification (RT-LAMP) is an interesting alternative to classical RT-qPCR testing. Collection of saliva is easy and non-invasive to collect, but it does contain disruptors and inhibitors of the enzymes in RT-qPCR and RT-LAMP. The aim of this study was to design a diagnostic method, based on colorimetric RT-LAMP for rapid detection of SARS-CoV-2 in saliva samples without prior isolation of viral RNA. We optimised the composition of the sample buffer, conditions for one-step colorimetric RT-LAMP reaction, and determined analytical sensitivity and specificity. In parallel, we also optimised the RT-qPCR method for the detection of viral RNA in saliva samples without the extraction of RNA. RT-LAMP diagnostic method was rapid, and it took only 50 minutes from sample collection to a readout of results. We found that the key component of the sample buffer was the RNA stabilisation buffer, and with the addition of RNase inhibitor (RNasecure) and chelating agent for removal of impurities (Chelex100), we further improved the quality of samples. RT-LAMP method targeting region N2 of SARS-CoV-2 genome was used to achieve analytical sensitivity of 615 molecules/ µl of the sample and an analytical specificity of 97.5 %. When the RT-LAMP method was tested in a real environment by testing saliva samples collected at the COVID entry point, we achieved 61 % sensitivity and 89 % specificity. With the RT-qPCR method, we achieved 95 % sensitivity and 100 % specificity on the same set of samples. We found that saliva is suitable for viral RNA detection even without viral RNA isolation, but the RT-LAMP method is less useful for this purpose than conventional RT-qPCR.