Recent advancements in neurobiology have provided increasing insights into the pathophysiology of neurodegenerative diseases, and opened doors to the development of molecular targeted therapies. Although many compounds showed positive results in animal studies, there is almost no drug for which the efficacy has been confirmed in clinical trials. This failure reflects a number of unsolved problems: limited knowledge of the exact pathways of neuron loss; safety and delivery issues of compounds; lack of established animal models that faithfully recapitulate human pathology; lack of validated, sensitive outcome measures; and limited tools to diagnose pre-symptomatic patients. To investigate the efficacy of potential disease modifying agents with limited financial and patient resources, the efficiency of both basic and clinical studies should be improved by integrated approaches. The reproduction of positive results from animal experiments that analyse the efficacy of compounds at symptomatic stages is needed to improve the credibility of preclinical studies. To effectuate proof of concept processes, novel designs of phase 2 clinical trials, such as the futility study, are being developed. Given the modest effects of molecular targeted therapies in human, it is necessary to explore clinical outcome measures that are resistant to variability, subjectivity and placebo. Furthermore, there is an increasing need for testing interventions before the onset of symptoms. Analyses of natural histories of biological and neurophysiological markers may provide indispensable information for designing such preventive trials. As it is now clear that conventional approaches are not necessarily appropriate for the development of molecular targeted therapies, both basic and clinical studies require conceptual innovation.