Eugene Volokh writes:

Martin Killias’s “International Correlations Between Gun Ownership
and Rates of Homicide and Suicide,” 148 Can. Med. Assoc. J. 1721, 1723-24
(1993), purported to show that “the proportions and the rates of homicide
and suicide committed with a gun as well as the overall rates committed by
any means were related to the rate of household gun ownership.”

Don Kates has just pointed out that a recent Killias work, Martin
Killias, John van Kesteren & Martin
Rindlisbacher, “Guns, Violent Crime, and Suicide in 21 Countries,” 43
Canadian J. of Criminology 429 (2001), concludes that there is no
statistically significant international correlation
between gun ownership
levels and total homicide, total suicide, total robbery, or total assault.
The study found some correlations between gun ownership and gun homicide,
suicide, and crime rates, but even those were present only as to some
categories of gun deaths and gun crimes. The abstract is here, but I read the entire
piece and that’s precisely what the article shows. A very singificant
development, it seems to me.

It would be an error to interpret this result as good evidence that
people substitute other means for homicide and suicide when firearms
are unavailable. As the authors note, firearms homicide and suicide
are a relativley small fraction of homicides and suicides in most of
the countries studied. This means that a study like this is not
likely to find good evidence either for or against the sustitution
hypothesis. Good evidence for sustitution would be significant
negative correlations between gun ownership and non-gun
suicide/homicide. Good evidence against substitution would be
significant positive correlations between gun ownership and total
suicide/homicide.