Mutational analyses have become crucial for therapeutic choices in patients with advanced lung cancer, colorectal cancer and melanoma. Short turnaround times for molecular analyses are necessary to match the patient's therapeutic management. Non-contributive molecular analyses may increase the delay in reaching a relevant mutational status. We attempted to identify criteria in samples associated with non-contributive molecular results to better anticipate them and select samples with contributive analyses. We compared several criteria such as cancer type, sample type, organ of origin and percentage of tumour cells between samples with non-contributive or contributive EGFR, KRAS, NRAS and BRAF mutation analyses. Among two sets of 3367 and 554 tumour samples analysed in 2015-2017 and 2018, respectively, 11.7% and 15.7% of sample analyses were non-contributive for at least one oncogene. Lung cancer and melanoma cancer subtypes [odds ratio (OR)=7.2], cytological (OR=1.8) or bone samples (OR=8.5) and a percentage of tumour cells ≤20% (OR=41.4) were significantly associated with non-contributive results. By combining these parameters in a scoring system, we were able to predict the contributive or non-contributive result of a molecular analysis with sensitivity and specificity higher than 80% in a validation set of samples. Predicting the contributive or non-contributive result of a molecular analysis is feasible in samples on the basis of simple features. A combination of these features could be used to better choose samples to analyse in order to reduce the rate of non-contributive molecular results and related treatment delays and costs in patients with advanced cancers.