Riluzole is the only approved medication that extends survival for patients with amyotrophic lateral sclerosis (ALS). While other potential neuroprotective agents have been evaluated in randomized clinical trials, none has shown unequivocal success and none has been approved by regulatory agencies. Few symptomatic therapies have been tested in ALS. Effectiveness for drugs with modest benefit can be established only through large phase III randomized clinical trials. With numerous potential agents but limited resources, priority should be given to agents that show promise in phase II trials before proceeding to evaluation in phase III trials. In this article, we review drug development in early phase ALS trials and introduce novel designs. First, to maximize the therapeutic potential of the test medication, we need to identify the highest dose that produces a tolerable level of side effects. Second, candidate treatments should be ranked by conducting randomized selection trials between competing new treatments. The selection paradigm adopts a statistical viewpoint different from the hypothesis testing framework in conventional trials. We exemplify this approach by describing a group-sequential selection design developed for a phase II, randomized, multicenter trial of two combination treatments in patients with ALS, and illustrate the sample size reduction from a conventional trial.