Periodic response of nonlinear oscillators is usually determined by ap
proximate methods. In the ''steady state'' type methods, first an appr
oximate solution for the steady state periodic response is determined,
and then the local stability of this solution is determined by analyz
ing the equation of motion linearized about this predicted ''solution'
'. An exact stability analysis of this linear variational equation can
provide erroneous stability type information about the approximate so
lutions. It is shown that a consistent stability type information abou
t these solutions can be obtained only when the linearized variational
equation is analyzed by approximate methods, and the level of accurac
y of this analysis is consistent with that of the approximate solution
s. It is demonstrated that these consistent stability results do not i
mply that the approximate solution is qualitatively correct. It is als
o shown that the difference between an approximate and the next higher
order stability analysis can be used to ''guess'' the role of higher
harmonics in the periodic response. This trial and error procedure can
be used to ensure the qualitatively correct and numerically accurate
nature of the approximate solutions and the corresponding stability an
alysis.