Though not a reptile, Gephyrostegus bohemicus (Jaeckel 1902) is where reptile evolution begins. It was an Upper Carboniferous (~310 mya) lizard-like tetrapod, or reptilomorph, from Nyran in the Czech Republic. Since some reptile fossils, like Westlothiana (~350 mya), are older than Gephyrostegus, it's clear we haven't found the oldest gephyrostegid yet.
Carroll (1970) thought that reptiles originated as tiny tetrapods and he was right! Cephalerpeton, the most primitive reptile was smaller than this specimen of Gephyrostegus. Even smaller was another specimen of Gephyrostegus. Read more about Gephyrostegus and the origin of reptiles here.
The reptile family tree begins with Gephyrostegus because, by definition, the Reptilia includes turtles, lizards, snakes, crocodilians and birds, their last common ancestor and all of its descendants. That last common ancestor (Cephalerpeton, according to the present study) was close to Gephyrostegus. In the present study, distinct from all prior studies, all known reptiles nest in two major branches with turtles and lizards on one branch and crodilians, birds and mammals on the other.
Gephyrostegus bohemicus itself was probably too large to produce embryos enclosed within an amniotic membrane, the defining character of the Amniota (= Reptilia in this study). However, Gephyrostegus watsoni is a more likely candidates.
Shortly after a very early sister to Cephalerpeton members of the Reptilia split into two clades represented at their bases by Westlothiana and Cephalerpeton. The new Lepidosauromorpha, are all those sharing a more recent common ancestor with living lizards, the Lepidosauria, than with Archosauria (Gauthier 1986). The new Archosauromorpha share more traits in common with living archosaurs (birds and crocs) than with Lepidosauria.
This primitive division has not been reported before because all prior studies were more limited in scope, without the benefit of an umbrella study like this one.
The more complete reptile family tree with archosauromorph and lepidosauromorph branches is found here. See a (slightly outdated) chronological phylogram of the same tree below. For those interested in obtaining the MacClade data file, click here and make your request.