<b>Introduction and background:</b> Artificial Neural Networks (ANN) are artificial intelligence tools used efficiently in the diagnosis of different diseases. The diagnostic performance of ANNs is unknown when compared with standardized questionnaires based on clinical information for the diagnosis of COPD. <b>Aims and Objectives:</b> To compare the diagnostic validity of two ANNs and the Lung Functional Questionnaire (LFQ) for the diagnosis of COPD. <b>Methods:</b> Diagnostic test study in patients undergoing lung function evaluation by spirometry and flow volume curve pre and post B2 in a hospital in Colombia. We evaluated the variables age, smoking, wheezing, dyspnea and expectoration through the LFQ questionnaire in its Spanish version in people older than 40 years, two types of ANNs were constructed. One Perceptron (ANNP) and one Radial Base Function (ANNRB). With 3 to 5 inputs, a hidden layer and a binary type output neuron, the diagnosis of COPD were made with a FEV1/CVF&lt;0.7 post B2 (ATS criteria). The sample size to contrast the hypothesis of two diagnostic tests was calculated, requiring a minimum of 1139 subjects. <b>Results:</b> 1514 subjects analyzed, COPD 16.8%, age: 64.7(SD: 12) years, 55.5% women, 47.3% with a history of smoking, Pack year: 6(R:0-188), AUC LFQ: 0.731(CI95%:0.698-7664), AUC (5 inputs) ANNP: 0.769(CI95%:0.739-0.799) and ANNRB: 0.777(CI95%:0.748-0.806), AUC (3 inputs) ANNP: 0.768(CI95%:0.760-0.817) and ANNRB: 0.777(CI95%:0.747-0.806), comparison ANNP vs LFQ p&lt;0.001, comparison ANNRB vs LFQ p&lt;0.001, comparison 5 inputs ANNP vs ANNRB p=0.377, comparison 3 inputs ANNP vs ANNRB p=0.071. <b>Conclusion:</b> The ANNs have a higher performance than LFQ questionnaire for the diagnosis of COPD.