Résumé : Aim: In elderly cancer patients, several studies assess the state of cognitive functions during treatment. The aim of this paper is to highlight the presence or absence of cognitive impairment at the beginning of cancer treatment. Procedure: One hundred thirty-two patients aged over 65 suffering from hematologic cancer were included. Tests and questionnaires were administered at the latest within 48 hours after the beginning of treatment: sociodemographic and socio-economic questionnaire, questionnaire on subjective and self-reported health, hetero-reported medical questionnaire, and Hospital Anxiety Depression Scale (HADS). A battery of four cognitive tests was then administered: digit span (DS), RL/RI 16, and Stroop Trail Making Test (TMT). Results: At the beginning of cancer treatment, first, elderly patients did not exhibit neither short-term memory impairment in DS, with notes of 9 standards for groups 1 (65-69 years) and 2 (70-74) and 8 for groups 3 (75-79) and 4 (80-89), nor long-term memory impairment in RL/RI 16. Secondly, lower performances of inhibition at half Stroop test have been found in patients with low (non corrected errors: percentile 5-10) and middle (non corrected errors: percentile < 5; reaction time: percentile 5-10) level of education to normalize at the end of the test. Nevertheless, no patient showed impaired flexibility. But the main result is that when patients have a lower level of education, the processing speed of information is slowed (Stroop and TMT). Conclusion: Patients aged over 65 suffering from hematologic cancers show no short- and long-term memory impairment but have a slower processing speed of information depending on the beginning of cancer treatment. © 2013 Springer-Verlag France.