Fingerprint plays an important role in forensic science, but latent print age estimation has been a difficult challenge. This paper proposes a model that is based on the fluorescence intensity of the amino acid solution reacted with 1,2-indanedione and other characteristics as well as the decision tree algorithm. First, a certain gradient concentration of 8 pure amino acid solutions is prepared, and fingerprints of 10 donators are collected to obtain samples with the age of 0, 3, 7, 14, 21, 30 days. The decision tree model is constructed based on the characteristic values of the sample which is reacted with 1,2-indanedione after being irradiated with a green laser and the fluorescence intensity observed with an orange filter, but the effect is not good. After excluding data less than 14 days age, the model is built again, the ACC Score reached 81%, and the ACC scores of the horizontal comparison with SVM (support vector machine) MLP (multilayer perception) model are 55% and 31%, respectively. The decision tree model is the optimal model. The model is tested with real fingerprint data greater than or equal to 14 days as the test set, and the ACC Score is 67%, which provides an idea for the judgment of latent print age.