Exploring metabolism in human tumors at the cellular level remains a challenge. The reduced form of metabolic cofactor NAD(P)H is one of the major intrinsic fluorescent components in tissues and a valuable indicator of cellular metabolic activity. Fluorescence lifetime imaging (FLIM) enables resolution of both the free and protein-bound fractions of this cofactor, and thus, high sensitivity detection of relative changes in the NAD(P)H-dependent metabolic pathways in real time. However, the clinical use of this technique is still very limited. The applications of metabolic FLIM could be usefully expanded by probing cellular metabolism in tissues ex vivo. For this, however, the development of appropriate tissue preservation protocols is required in order to maintain the optical metabolic characteristics in the ex vivo sample in a state similar to those of the tumor in vivo. Using mouse tumor models of different histological types-colorectal cancer, lung carcinoma and melanoma-we tested eight different methods of tissue handling by comparing NAD(P)H fluorescence decay parameters ex vivo and in vivo as measured with two-photon excited FLIM microscopy. It was found that the samples placed in 10% BSA on ice immediately after excision maintained the same fluorescence lifetimes and free/bound ratios as measured in vivo for at least 3 hours. This protocol was subsequently used for metabolic assessments in fresh postoperative samples from colorectal cancer patients. A high degree of inter- and intra-tumor heterogeneity with a trend to a more oxidative metabolism was detected in T3 colorectal tumors in comparison with normal tumor-distant colon samples. These results suggest that the methodology developed on the basis of FLIM of NAD(P)H in tissues ex vivo show promise for interrogating the metabolic state of patients' tumors.