The proton magnetic resonance spectrum suffers from limited dispersion of signals along the chemical shift axis, leading to substantial overlap of signals from metabolites present at potentially detectable levels. Spectral editing addresses this overlap problem by reducing the number of signals in the spectrum, resolving particular signals of interest. Low-concentration metabolites with coupled spin systems that benefit from edited detection include gamma aminobutyric acid (GABA), glutathione (GSH), lactate (Lac), and N-acetyl aspartyl glutamate (NAAG). The most widely used editing approach is J-difference editing, in which two experiments are acquired that differ in their treatment of a known scalar coupling within a molecule of interest. Frequency-selective pulses can be applied in one experiment, so as to refocus evolution of the coupling, and not in the other, so that the coupling evolves throughout the echo time. The difference between these two experiment will be a spectrum which only contains signals that are impacted by the editing pulses. The accurate subtraction of unedited signals is the main special processing requirement for edited data. In this presentation, we will cover the major steps required for the analysis of edited spectra, which include the standard steps used for all spectroscopy (Fourier transformation, windowing/filtering, integration/fitting) and some steps that are specifically required by editing (subtraction, frequency-and-phase correction of time-resolved data). Data acquired in J-difference-edited experiments are stored in the time domain, without signal averaging, as an interleaved series of editing-on and editing-off. In general, such data require a number of processing steps to yield spectra and ultimately concentration estimates: down-sampling; coil combination; windowing; Fourier transformation; frequency-and-phase correction of individual transients; time averaging of editing-off and editing-on spectra; quantification.We will briefly discuss the steps applied to all MRS data, with emphasis on the steps whose parameters and order differ for edited data. The post-processing step that is required for difference-edited data is frequency and phase correction of individual transients. We will review the literature that shows such processing is necessary and demonstrate that the application of inappropriate methods can increase rather than decrease the presence of artefacts in spectra. Both time-domain methods, such as Spectral Registration, and frequency-domain methods, such as modeling of the Cr or NAA signal, will be considered. We will show that that method of choice will depend on the data being corrected, because both the direct saturating effect of editing pulses and the modulation of the edited signal can interfere with the underlying assumption that all signals are the same to within a frequency/phase correction factor. For example, both Creatine- and NAA-based corrections fail for GABA editing, for different reasons, whereas time-domain methods struggle for glutathione editing. We will also discuss the interpretation of frequency traces that arise from such corrections, including identifying motion and field drift. The choice of experimental acquisition parameters will be influenced by the post-processing steps that are used. For example, complete experimental water suppression is no necessary when an additional order of magnitude arise from on-off subtraction, and the complete removal of the water signal means it cannot be used to determine frequency and phase corrections. Similarly, narrowband water suppression can interact unpredictably with B0 field changes arising from drift or subject motion, restricting the use of water-based, or time-domain correction methods. As with standard MRS data, methods of quantification vary from baseline correcton and integration, through simple curve-fitting to linear-combination modelling. Most tools developed for the analysis of standard MRS data, including LCModel and Tarquin, can be applied (with care) to the quantification of edited spectra. We will also discuss tools specifically developed for the analysis of edited data, including Gannet. The main aim of the edited experiment is a simplification of the spectrum, and quantification becomes a simpler problem accordingly. However, the separation of signals with editing is rarely complete. For example, common GABA editing protocols co-edit substantial amounts of macromolecular signal. Glutathione editing coedits aspartyl signals that complicate modelling. Finally, we will consider quantification beyond modelling, including measuring in vivo relaxation parameters for edited metabolites, and consideration of editing efficiency. The incorporation of voxel segmentation data (grey matter, white matter and cerebrospinal fluid fractions) into quantification equations will also be discussed.