It may be possible to describe and diagnose 4 lung diseases of the group CNSLD by laboratory data only, by computer, apart and separated from physician’s diagnosis. Recording of concentration curves of expired gases yield 4 lung parameter values: He_1•5, MHe, RHe, RCO₂. These, together with 8 other parameter values: FEVC, FEV₁ % FEVC, M₁ (= FEV₁%FIV₁%), MVV₃₀/ FIVC, FRC % TLC, RV % TLC, DLCO/body surface area, reversibility of FEV₁, and 3 administrative data: age, season and sex, form 15 ‘symptoms’, used together with ‘a priori probability’ to compute diagnosis of (A) asthma bronchiale, (B) bronchitis asthmatica, (C) bronchitis chronica and (E) emphysema pulmonum, from 703 patients in one clinic. The numerical values of the symptoms are grouped in symptom classes. By using chi-square test, Wilcoxon test, and T (=Δ/SE/Δ) test with and without classes, each disease (A, B, C or E) significantly differed in at least 7, 8, 8 and 10 symptoms, respectively, (on the average 12) from the other and in at least 11 symptoms from normal condition. Rank lists for discriminative properties of the symptoms show the importance of the gas concentration curve parameters RCO₂, RHe, MHe and He_1•5 and also of the parameters: reversibility and M₁.