Background <p>The formulae for calculating intraocular lens (IOL) power have undergone considerable development. In recent years in particular, novel approaches based on artificial intelligence (AI) have been incorporated into the calculation models.</p> Objective <p>The aim is to discuss both classical and new principles of IOL calculation, to present the current classification, to explain new approaches and to compare the performance of the formulae based on current reviews and meta-analyses.</p> Material and methods <p>This article is a&#xa0;narrative review without any claim to completeness. The literature selection was based on relevant publications identified through medical databases and specialist journals. Initial publications on the formulae and articles on classification systems were identified and analyzed, followed by an evaluation of current reviews and meta-analyses to compare the performance capabilities.</p> Results <p>The spectrum of IOL formulae ranges from historical refraction-based and regression-based approaches, classical vergence and ray tracing to modern AI-based methods. Many of the contemporary formulae used integrate multiple methodologies. In recent years there has been a&#xa0;clear trend towards reducing prediction errors with the advent of newer formulae, particularly in short eyes.</p> Conclusion <p>Modern IOL formulae are preferable to traditional ones and outdated guidelines regarding formula selection should be updated. Many modern formulae are universally applicable and yield similarly good results. Further developments will reduce the number of outliers but will likely never eliminate them entirely.</p>