The aim of this study was to evaluate genetic variability in the Turkish Arab horse population using pedigree information. This study is the first detailed pedigree analysis of the breed in Turkey. Pedigree data were collected from the National Studbook. The pedigree data for 23 668 horses, born between 1904 and 2014, were used in the analysis. From this data set, a reference population (RP) of 14 838 animals symbolising the last generation was defined. Demographic parameters, the inbreeding level (F), the average relatedness (AR), the effective population size (Ne), the effective number of founders (fe), the effective number of ancestors (fa) and the number of founder genome equivalents (fg) were calculated for the population. The average generation interval for the RP was 12.2±4.6 years, whereas the calculated pedigree completeness levels were 98.2%, 96.6% and 95.0% for the first, second and third known generations. The mean equivalent generations (t), the average complete generations and the mean maximum generations for the RP were 7.8, 5.4 and 12.2, respectively, whereas the mean F and AR were 4.6% and 9.5% for the RP. The rate of inbred animals was 94.2% for the RP, whereas the number of founders, the number of ancestors and the fe, fa and fg were 342, 223, 40, 22 and 9.6 for the RP. The large differences observed between fe, and the number of founders demonstrates that genetic diversity decreased between the founder and the RP. Contribution of the 14 most influential founder to the RP was 50.0%, whereas just eight ancestral horses can account for 50% of the genetic variability. Ne estimated via an individual increase in inbreeding per generation ( $$ \hskip3pt{\bar{\hskip-2ptN}_{{\rm e}} } $$ ), and paired increase in coancestry $$\left(\! \hskip3pt{\bar{\hskip-2ptN}_{{ eC}} } \right)$$ , were 74.4±3.9 and 73.5±0.58, respectively. The inbreeding increases with the pedigree knowledge. In addition, the decrease in inbreeding in last years is more noticeable.