Suicidal behaviour is the most common psychiatric emergency. A large proportion of suicidal behaviour can be prevented, particularly in cases associated with mental disorders. Early recognition of suicidality and reliable evaluation of suicide risk are crucial for the clinical prevention of suicide. Evaluation of suicidal risk involves assessment of suicidal intent, previous suicide attempts, underlying psychiatric disorders, the patients’ personality, the social network, and suicide in the family or among acquaintances as well as other well-known risk factors. Suicide risk assessment should take place on several levels and relate to the patient, the family and social network but also to the availability of treatment, rehabilitation and prevention resources in the community. As suicide risk fluctuates within a short period of time, it is important to repeat the suicide risk assessment over time in an emphatic and not mechanistic way. The suicidal person may mislead both family members and hospital staff, giving a false sense of independence and of being able to manage without the help of others. Although extreme ambivalence to living or dying is often strongly expressed by the suicidal individual, it is not seldom missed by others. If observed in the diagnostic and treatment process, dialogue and reflection on such ambivalence can be used to motivate the patient for treatment and to prevent suicide. If ambivalence and suicidal communications go undiscovered, the treatment process and the life of the patient can be endangered. Today, several measurement tools of suicide risk exist, including psychometric and biological measurements. Some of these tools have been extensively studied and measures of their sensitivity and specificity have been estimated. This allows for the formulation of an approximate probability that a suicidal event might happen in the future. However, the low precision of the predictions make these tools insufficient from the clinical perspective and they contribute very little information that is not already gained in a standard clinical interview. Psychiatrists and other mental health professionals have always longed for reliable and precise tools to predict suicidal behavior, which could support their clinical practice, allow them to concentrate resources on patients that really need them, and backup their clinical judgement, in case of eventual legal problems. In order to be useful, however, the approximate probability that a suicidal event might happen in the future is not sufficient to significantly change clinical routines and practices. These should rely on the available evidence base and always consider the safety of the patient as paramount.