Rate transient analysis (RTA) is an important reservoir engineering tool used to extract reservoir and stimulation information from flow rates and pressures obtained from producing wells. Adaptation of analytical RTA methods for this purpose has been slow for coalbed methane (CBM) reservoirs because of the complexities of multi-phase flow, non-static absolute permeability, desorption, among others. Some progress has been made historically for high permeability CBM reservoirs, where depletion (boundary-dominated flow) occurs quickly, however low permeability CBM reservoirs are currently being evaluated internationally (e.g. China), which brings forward additional challenges. For example, in some instances, low permeability, undersaturated CBM wells exhibit long periods of transient flow, during which the transition from single-phase flow of water to two-phase flow of gas and water occurs. This behavior has not been dealt with rigorously in the development of analytical RTA methods. It is the purpose of the current work to extend RTA to apply to low-permeability, undersaturated CBM wells exhibiting this behavior. The horizontal CBM well studied in this work exhibits two periods of apparent transient linear flow separated by a transition, as determined from diagnostic analysis of water production data using a log-log plot of rate-normalized pressure (RNP) versus time. It is hypothesized that the well actually exhibits transient linear flow throughout the entire studied production period (supported by reservoir simulation) and that productivity changes leading to the apparent transition in flow regime are caused by the combined effects of relative permeability changes after desorption pressure is reached, and (possibly) fines migration and stress-dependent permeability. In order to extend the early single-phase linear flow analysis of this well (occurring above desorption pressure) to the multi-phase period (below desorption pressure), the RNP half-slope trend (linear flow) established during the single-phase flow period is extrapolated through the rest of the production period. An RNP ratio (RNPR) is then created by calculating the ratio of the extrapolated (corrected) RNP and the RNP of the uncorrected field data. The RNPR can then serve to correct the production data affected by multi-phase flow (and possibly fines migration and stress dependent permeability) and hence extend the amount of data that can be used for production analysis. Further, the RNPR squared, combined with water saturation calculations, may be used to derive water relative permeability curves from the transient data. These concepts are tested using both numerical simulation and field data. The resulting field-derived water relative permeability curve can be used as a solid starting point for simulation history-matching, although it covers a limited range of water saturations. This work should be of interest to professional engineers and geoscientists involved in CBM reservoir characterization efforts. Future work will determine if these concepts are applicable to other low-permeability reservoirs exhibiting the transition from single- to multi-phase flow during the transient linear flow period, such as tight oil and gas condensate reservoirs. (C) 2015 Elsevier B.V. All rights reserved.