Background/aims Serial Scheimpflug corneal tomography to monitor the progression of keratoconus has become standard practice in most countries where corneal cross-linking is available. The tomographic definitions of progression are, however, poorly defined. The aims of this study were: (a) to estimate the 95% limits of intraobserver and interobserver agreement of corneal shape parameters on Pentacam in patients with keratoconus and (b) to investigate whether these limits of agreement varied according to disease severity.

Methods 96 adult patients with keratoconus and no corneal scarring or history of previous surgery were recruited from a corneal clinic in a tertiary ophthalmology hospital. One eye of each subject was scanned twice by each of the two observers with the Pentacam HR. 95% limits of intraobserver and interobserver agreement for K1, K2, Kmax and corneal thickness at the thinnest corneal location (TCT) were calculated.

Results Reproducibility of keratometry measures was better for early keratoconus than advanced keratoconus. In patients of Pentacam-derived Krumeich stage 1 or 2, the 95% limits of interobserver agreement for Kmax were from −0.90 to 1.01. In patients of Pentacam-derived Krumeich stage >2, the 95% limits of interobserver agreement for Kmax were from −3.71 to 3.86.

Conclusions Keratometric measurements on Pentacam HR are less reproducible in advanced keratoconus than in early keratoconus. In patients of Pentacam-derived Krumeich stage 1 or 2, an increase in K1, K2 or Kmax of more than 1 dioptre is likely to represent the real change in the corneal shape.