In this consecutive sample of arsonists, 22.4% were recidivists, which is similar to the results of some fire-setting studies undertaken among the forensic population [7, 11]. The most important diagnostic categories of arson recidivists were personality disorders, psychosis and mental retardation, often with comorbid alcoholism. Nearly 70% of the arsonists were under acute alcohol intoxication during the index crime; these observations were similar to those reported in previous fire-setting studies [7–9].
The mentally retarded arson recidivists
18% of recidivists were mentally retarded persons and they were almost all pure arsonists. Originally, pathological fire-setting was considered as a phenomenon perpetrated by physically disabled or mentally retarded pubescent girls who had abnormal psychosexual development . Later, due to methodological difficulties, there have been conflicting opinions of whether people with mental retardation are over- or underrepresented in the arsonist population . In a recent prospective follow-up of 61 offenders with intellectual disability, both sex offence and arson were overrepresented offence types and as many as 21.3% of the total population had a history of fire-setting . In contrast to our results, most of the above-mentioned 61 persons with intellectual disability had committed also other offences. A possible explanation for this discrepancy might be that the above mentioned study group included single-setters with an IQ below 80, whereas in our study all the subjects were repeaters and the IQ cutoff point was 70 or below, which generally indicates a significantly impaired intellectual functioning. As far as we know, no previous studies have focused particularly on arson recidivists in this population. Murphy and Clare  reported that the three most frequently identified antecedent emotions/events before fire-setting in mentally retarded patients were, in rank order, angry feelings, not being listened to/ attended to, and feeling sad/ depressed. Stereotypical behaviour is frequently found in deeply mentally retarded patients , which most of the persons in the present study were not. However, overrepresentation of mentally retarded persons as pure arsonists ma be explained by the idea that they represent persons who express their feelings of anger, frustration and sadness through specific, repetitious action models (in these cases arson), but who do not otherwise express interest in criminal activities.
The psychotic arson recidivists
20 % of recidivists were psychotic persons and, like mentally disordered patients, they were mostly pure arsonists. A higher than expected prevalence of psychotic disorders among fire-setters was already reported over 60 years ago . Motives of patients with schizophrenia have been reported to be often similar to those observed in non-psychotic persons, with hatred and revenge as the common underlying factors in arson . However, delusions have also been regarded as motives or reasons . In a very large sample of fire-setters as many as 40% of paranoid schizophrenics repeated fire-setting . A large proportion of recidivists set fires to property that is unrelated to them in any way and they also seem to set fires on a very sporadic basis, usually within the context of situational crises . In our sample, most psychotic persons were pure arsonists with no other criminal activities, suggesting, that among psychotic fire-setters there may be different subgroups with various underlying psychological mechanisms leading to fire-setting behaviour. In single-setters the motives may be more instrumental, whereas in recidivists fire-setting is probably a function of flagrant thought disturbance, precipitated by command hallucinations and delusions.
The arson recidivists with personality disorders
Antisocial personality disorder was the most common personality disorder in the present sample (22 % of recidivists) and, in fact, some of the best predictors of recidivist fire setting are impulsive characteristics . Deliberate fire-setting is one of the diagnostic criteria in conduct disorder, the childhood and adolescent predecessor of adult antisocial personality disorder , and self-reported fire- setting is strongly associated with extreme antisocial behaviour in young community adolescents . Recently, adolescent fire-setters were described as aggressive but on the other hand shy and rejected by their peers . All recidivists with antisocial personality disorder were intoxicated during the index crime. In a prospective follow-up study, low central nervous system 5-hydroxyindole acetic acid (CNS 5-HIAA) and 3-methoxy-4-hydrophenylglycol (MHPG) concentrations, typically associated with impulsive antisocial personality and Cloninger type II alcoholism, were associated with both violent recidivist criminal offences and fire-setting . In the present sample all the antisocial fire-setters were nonpure arsonists, and the arson were often aimed at revenge. The underlying motives for these impulsive acts were typically hatred and rage, which increased in severity along with acute substance intoxication. Offenders with severe personality disorder typically begin their criminal careers early and their criminal records are extended. As criminals, their acts appear to be heterogeneous and fire-setting reflects their criminal and unempathic natures . In a study by Repo et al.  among arsonists, overall lifetime criminal recidivism was primarily associated with antisocial personality and alcohol dependence.
The arson recidivists with pyromania
Pyromania is regarded as an impulse-control disorder, but controversy continues over whether the condition should be classified under the impulse-control disorders or, indeed, whether it comprises a separate entity at all. Pyromania was already mentioned in DSM-I, omitted from the DSM-II, but then acknowledged again in the DSM-III . In the DSM-IV-TR a wide range of exclusion criteria such as psychotic states, dementia, mental retardation, monetary gain and political ideologies are presented. The exclusion criteria cover also other criminal acts, rage, revenge and acute intoxication. The phenomenon should also not be observed as pyromania if it is better explained as conduct disorder, antisocial personality disorder or manic state. In our sample, after exclusion of persons with psychosis, mental retardation, organic brain syndrome and antisocial personality disorder, 12 of the arson recidivists fulfilled the inclusion criteria of pyromania. They expressed tension or affective arousal before the act, attraction to and interest in fire, as well as pleasure and release afterwards. However, nearly all these persons were intoxicated during the fire-setting, and they typically mentioned that the tension as well as the affective arousal increased when they were consuming alcohol. Since substance intoxication is one of the exclusion criteria, only three persons with true pyromania were found in the present sample. They all were pure arsonists and, interestingly, all these three men worked as volunteer firefighters and also expressed special interest in fire in other ways as arsonists. In conclusion, using the present strict exclusion criteria for pyromania, the disorder must be regarded as an extremely rare phenomenon as also shown in previous studies [10, 11]. The question of substance intoxication as an exclusion criterion should especially be reconsidered.
In Finland, an average of about 500–600 arson attempts are made and 100 offenders convicted of arson each year. The proportion of individuals who undergo a forensic psychiatric evaluation compared with all arsonists suspected by the police was estimated to be only 10% . So, the present sample, as well as other previous pretrial samples reported in the literature, is not representative of arsonists in general. The diagnoses were based on several clinical interviews, psychological tests and observation during the two month examination periods. Finnish forensic psychiatric examination statement traditionally includes a paragraph involving a review of the person's previous official criminal history. This information instead of authentic criminal records was used in dividing arsonists into different subpopulations. The strength of the study is that we were able to collect a consecutive sample of male fire-setting recidivists over a period of more than two decades. However, further research is still needed to clarify the important question of fire-setting recidivism.