Petrographic changes during coalification of three peat samples were investigated by artificial coalification experiments using a semi-open reactor system. Compressions of from 83 to 88% produced dark brown to black, shiny, flattened pellets exhibiting microscopic banding. Overall amounts of compression inversely correlated with the framework to matrix ratios ( F M ); although, certain tissue fragments remained relatively uncompressed due to the presence of tanniniferous cell fillings and also, hypothetically, to the presence of colorless plant secretions or gelified wall materials that were present in the cells but only recognizable after a change in color during coalification. The most distinct microscopic banding (longest and widest bands) developed in samples of the dome-formed peat facies, which had the highest framework to matrix ratios of any type and the greatest contents of surface litter composing the framework (i.e. highest S N ratio). The planar, root-dominated, Rhizophora facies showed the greatest change in microbanding character during coalification; in that, the numerous lens-like bands produced by the compressed roots were very distinct at the 60°/2100 psi (121.4 kg/cm2) stage but nearly disappeared at the 175°/5000 psi (289.1 kg/cm2)) stage. This is explained as a result of differences in rates of coalification of the different telinitic precursors in the roots (multilinear vitrinization) and could explain why roots are significant components of many modern peats (and ancient coal ball concretions) but are difficult to recognize in bituminous coals. Similarly, separation of cuticles from leaves in modern peat-forming environments, due to irreversible shriveling and subsequent enhanced microbiological decay, partially explains why leaf remains are also difficult to recognize in most coals, even those with abundant cutinites. Microcracks developed in all pellets, with dominant orientations of cracks being either perpendicular to induced banding or horizontal to it. Minute, vertical microcracks (interpreted as ‘dewatering’ features) were observed primarily in gelinitic bands, especially in the most fire-prone, Cladium coal facies; but, these rarely extended into juxtaposed microlithotypes as did most of the larger vertical cracks. Horizontal cracks tended to form at microband boundaries. Almost all humotelinitic and humocollinitic precursors showed some changes in color during artificial coalification, with humotelinites exhibiting less or slower changes than humocolinites or humodetrinites (finer-grained matrix material). Additionally, nearly all dark brown, degraded tissues in the peats became dark orange or red during coalification; but, many of these tissue masses were observed to contain significant amounts of minute inertinite derived from microorganisms (‘pseudomicrinite’) that only became prominent after coalification and subsequent change in color of surrounding macerals. Fusinite precursors were only found in the fire-prone Cladium facies; and, although, the coalified, doming-generated, Cyrilla facies exhibited evidence of significant drying events (i.e. occasional bands or parts of bands containing significant pseudomicrinites, macrinites and sclerotinites), it was dominated by huminitic (vitrinitic) components. Liptinites were most common in the coalified, dome-formed, Cyrilla facies and displayed a slight qualitative increase in red/green fluorescence during coalification and some possible flow of components into nearby cracks. Changes in mean random reflectance of huminitic macerals, although still requiring more data and only measured to the 60°C/2100 psi (121.4 kg/cm2) level, indicated increasing directions of change during artificial coalification, with humocollinitic macerals progressing from an average of about 0.21% to 0.32% and humotelinitic macerals (which consistently had lower inherent reflectances in the peats) progressing from 0.15 to 0.25%. The above petrographic results, along with previously reported chemical results, all suggest that the methods that we have used in these experiments have produced changes that might reasonably be expected to occur during natural coalification (despite the fact that we have speeded up the process). Although more work needs to be done to verify and refine these results and to establish a correlation between these artificial ‘coalification steps’ and true coalification, the new observations and conclusions from these studies might still be helpful in constructing models to predict or interpret coal seam characteristics and to establish the timing and release-potential of gaseous or liquid hydrocarbons from coals.