Previous research carried out in our laboratories has shown that all the ke
y parameters of electrothermal atomic absorption spectrometry, gas phase te
mperature, intensity of the probing beam and number density of absorbing sp
ecies, are generally highly non-uniform over the absorption volume. Further
, it was shown that, when using conventional detection systems based on a p
hotomultiplier tube or a photodiode that can only detect radiation spatiall
y integrated over their working area, absorbance measurements are subject t
o photometric errors when the absorbing layer is spatially non-uniform. Thi
s error is eliminated when using spatially-resolved detection of transmitte
d intensities with a linear solid state detector (photodiode array, linear
CCD). The photometric error of the conventional detection systems does resu
lt in an analytical error, if analyte distributions in the absorption volum
e produced from an aqueous standard solution and the unknown sample are dif
ferent. Such a differing distribution could be created under the influence
of the sample matrix on the analyte gas phase distribution. An atomic absor
ption spectrometer is described in the paper that allows spatially and temp
orally resolved detection of both specific and non-specific absorbances. Th
e effect of sample matrix on the analyte gas phase distribution is investig
ated when atomizing some environmental samples and, for the first time, the
results of spatially-resolved atomic absorption determination of cadmium a
nd lead in these samples are presented. It is shown that the influence of t
he matrix on the analyte distribution is significant, resulting in a signif
icant analytical error. By avoiding such errors, the benefits of atomic abs
orption analysis with spatial resolution over conventional AAS are directly
demonstrated.