Railway transport is considered one of the most reliable, comfortable and safest modes of travel for both freight and passengers. Rail infrastructure assets (such as tracks, bridges, earthworks, tunnels and drainage systems) must be inspected and maintained on a regular basis in order to ensure that transport services are delivered in compliance with contractual and legal obligations. The maintenance of railway track structures is preventive in nature and includes the repair or replacement of certain components at pre-determined time intervals (in terms of years of operation) and/or usage rates (in terms of gross tonnage). Maintenance actions such as grinding and stone-blowing either restore the track profile to its original condition, i.e., “as good as new (AGAN)”, leave the track in almost the same condition as it was in prior to the inspection, i.e., “as bad as old (ABAO)”, or restore the track condition to a state somewhere between AGAN and ABAO, i.e., the so-called imperfect maintenance. The effect of an imperfect maintenance is often characterized by two classes of models, namely, failure-intensity reduction and age-reduction. However, the impact of imperfect repair on assets’ usage has not yet been addressed in the literature. In this paper, a usage-based imperfect preventive maintenance (PM) planning model is proposed for railway track superstructures, where the effect of an imperfect maintenance is described by a reduced amount of total accumulated million gross tons (MGT) passed over the rail line. A constrained nonlinear programming model is formulated to optimize the maintenance interval (i.e., usage rate between consecutive PMs) and the degree (quality) of repair actions. The total mean maintenance cost for a Weibull failure distribution model is derived and, then, the conditions required to make PM actions beneficial are discussed. A numerical case example is provided to show the effectiveness of the proposed PM planning method over the track renewal and replacement policy.