In the area of lead optimization for potential CNS-active drugs in medicinal chemistry, there is a great need for experimental methodologies that can generate data relevant to estimates of free (unbound) drug exposure within the CNS. The methods chosen have to be efficient and have to measure a pharmacologically relevant entity. The lack of methods for generating such data is probably linked with the lack of successful lead optimization strategies within CNS drug discovery. This article evaluates available methods for estimating drug delivery to the brain, and discusses the relevance of the methods from the perspective of CNS exposure to free drug. It is suggested that the extent of drug delivery is the most important investigative parameter, since permeability (rate of transfer) can vary within a relatively wide range and still allow effects within the CNS. Following this suggestion would shift the focus from the current way of thinking and could lead to the development of less lipophilic compounds than are currently being investigated. It is concluded that an extensive collection of quality data on brain drug delivery, transporter affinities and in vivo behavior is urgently required so as to be able to build relevant predictive in vitro and in silico models for the future. These models need to be much more focused on the asymmetry of active transport across the BBB than on permeability data.