Fluorescence spectroscopy is a promising tool for the characterisation of biological tissues and shows the potential for in vivo clinical diagnosis in many diseases. It exploits the inherent photo-physical properties of a number of endogenous molecules. Fluorescence lifetime measurements of these fluorophores such as NADH and flavoproteins present an opportunity to discern functional information concerning myocardial energetics without issues of toxicity or systemic effects associated with the introduction of exogenous compounds. Additionally, autofluorescence from extra-cellular matrix molecules e.g. collagen may provide information on structural changes to the heart. Measurements of fluorescence lifetime are independent of fluorophore concentration, excitation intensity, sample attenuation and other experimental artefacts and can also report on changes to the fluorophore microenvironment e.g. pH and protein binding state. Thus we are interested to develop autofluorescence lifetime (AFL)-based techniques as a myocardial "optical biopsy". We report the application of a custom fibre-optic probe-based time-resolved spectrofluorometer utilising time-correlated single photon counting that we developed to characterise the autofluorescence signatures associated with the histological, morphological, metabolic and functional changes in myocardial tissue in health and disease states. We studied an in vivo rat left anterior descending coronary artery ligation heart failure (MI-HF) model 16 weeks post-infarction, which is a model well characterised in our institution. We observed stable AFL signals across age-matched control (AMC) hearts (n = 6) in different anatomical locations. We observed significant differences in AFL signals between MI-HF (n = 6) and AMC (n = 6) in infarcted regions: left ventricle (LV) anterior wall (p < 0.0001) and "border zone" region (p < 0.0001). We also observed significant differences between MI-HF and AMC in remodelled regions distant to the infarct: LV posterior wall (p < 0.01) and right ventricle (RV) (p < 0.001). Application of principal component analysis and linear discriminant analysis to the data facilitated development of a diagnostic algorithm to differentiate MI-HF and AMC in a given anatomical location with a high degree of accuracy both in infarcted regions (specificity 100%, sensitivity 100%) and even in remote regions e.g. RV (specificity 96%, sensitivity 89%). This represents, to the best of our knowledge, the first in vivo application of time-resolved fluorescence spectroscopy to the study of the heart. This technique shows promise as a diagnostic tool and could be readily and rapidly translatable into clinical cardiology practice.