Autophagy has been implicated in a range of disorders and hence is of major interest. However, imaging autophagy in real time has been hampered by lack of suitable markers. We have compared the potential of monodansylcadaverine, widely used as an autophagosomal marker, and the Atg8 homologue LC3, to follow autophagy by fluorescence microscopy whilst labelling late endosomes and lysosomes simultaneously using EGFP-CD63. Monodansylcadaverine labelled only acidic CD63-positive compartments in response to a range of autophagic inducers in various live or post-fixed cells, staining being identical in atg5(+/+) and atg5(-/-) MEFs in which autophagosome formation is disabled. Monodansylcadaverine staining was essentially indistinguishable from that of LysoTracker Red, LAMP-1 or LAMP-2. In contrast, 60-90% of EGFP-LC3-positive punctate organelles did not colocalise with LAMP-1/LAMP-2/CD63 and were monodansylcadaverine-negative while EGFP-LC3 puncta that did colocalise with LAMP-1/LAMP-2/CD63 were also monodansylcadaverine-positive. Hence monodansylcadaverine is no different from other markers of acidic compartments and it cannot be used to follow autophagosome formation. In contrast, fusion of mRFP-LC3-labelled autophagosomes with EGFP-CD63-positive endosomes and lysosomes and sequestration of dsRed-labelled mitochondria by EGFP-LC3- and EGFP-CD63-positive compartments could be visualized in real time. Moreover, transition of EGFP-LC3-I (45 kDa) to EGFP-LC3-II (43 kDa)-traced by immunoblotting and verified by [(3)H]ethanolamine labelling-revealed novel insights into the dynamics of autophagosome homeostasis, including the rapid activation of autophagy by the apoptotic inducer staurosporine prior to apoptosis proper. Use of fluorescent LC3 and a counter-fluorescent endosomal/lysosomal protein clearly allows the entire autophagic process to be followed by live cell imaging with high fidelity.