Autofluorescence bronchoscopy (AFB) has been shown to be a highly sensitive tool for the detection of early endobronchial cancers. When excited with blue-violet light, early neoplasia in the bronchi tend to show a decrease of autofluorescence in the green region of the spectrum and a relatively smaller decrease in the red region of the spectrum. Superposing the green foreground image and the red background image creates the resultant autofluorescence image. Our aim was to investigate whether the addition of backscattered red light to the tissue autofluorescence signal could improve the contrast between healthy and diseased tissue. We have performed a clinical study involving 41 lung cancers using modified autofluorescence bronchoscopy systems. The lesions were examined sequentially with conventional violet autofluorescence excitation (430 nm+/-30 nm) and violet autofluorescence excitation plus backscattered red light (430 nm+/-40 nm plus 665 nm+/-15 nm). The contrast between (pre-)neoplastic and healthy tissue was quantified with off-line image analysis. We observed a 2.7 times higher contrast when backscattered red light was added to the violet excitation. In addition, the image quality was improved in terms of the signal-to-noise ratio (SNR) with this spectral design.