Check out this really nice resource from The Allen Brain Institute! http://celltypes.brain-map.org/ 

I have my own question. Are you sure that neurons can unambiguously be classified as inhibitory or excitatory?  For instance, photoreceptor cells in the retina make synaptic contacts with two sorts of bipolar cells. One type of the bipolars is depolarized, and the other, hyperpolarized by light. So is the rod or cone inhibitory or excitatory? Also notice that photoreceptors themselves are hyperpolarized by light.

To my knowledge there is no unequivocal way to differentiate between excitatory and inhibitor neurons based on their respective electrophysiological properties. As Victor I Govardovskii has pointed out, above, in the case of some neurons the question is moot, since they are at once excitatory and inhibitory simultaneously.

Action potential waveform width at half max (aka AP half width) and membrane time constant are two good ephys parameters that can distinguish between broad classes of excitatory vs. inhibitory neurons.  In many cases fast afterhyperpolarization amplitude is pretty discriminatory.  The Allen Institute Cell Types database has data to that effect (for visual cortex neurons), although the parameters displayed have some obscure names, such as upstroke/downstroke ratio, which for the most part captures similar info to AP half width, or fast AP trough, which is basically fast afterhyperpolarization amplitude.  There are always going to be exceptions to the rule, but those are the uncommon cases.

While AP half-width (measured in whole-cell configuration) does well at identifying one class of inhibitory cortical neuron (fast-spiking, parvalbumin-positive), I must caution against using AP half-width to broadly distinguish excitatory vs. inhibitory neurons.

By recording from multiple classes of inhibitory and excitatory neurons within a single cortical layer (two excitatory types and two inhibitory types in L5B in motor cortex), we showed that a major class of inhibitory interneuron (low-threshold spiking aka LTS aka somatostatin-positive) has an AP half-width intermediate between two types of excitatory pyramidal neurons (corticospinal and crossed-corticostriatal). 

See Fig. 3 (attached) from our publication:

Suter, B. A., Migliore, M. & Shepherd, G. M. G. Intrinsic electrophysiology of mouse corticospinal neurons: a Class-Specific triad of Spike-Related properties. Cerebral Cortex 23, 1965-1977 (2013).

Thus, studies that split neurons into "putative inhibitory" and "putative excitatory" on the basis of AP half-width will tend to erroneously group non-fast-spiking inhibitory neurons into the "putative excitatory" population. Granted, these are less numerous than their excitatory neighbours, which may somewhat minimize the severity of such errors.

In some cases it may be possible to distinguish cell-type by considering a combination of intrinsic parameters (as described above), however care must be taken to employ the same recording conditions (temperature, internal solution, cortical area and layer) that were considered for the initial classification.

Article Intrinsic Electrophysiology of Mouse Corticospinal Neurons: ...

Inhibitory cells tend to fire more quickly either spontaneously or in response to current injection. They are also 'tighter' electrically and therefore the AP amplitudes are often shorter. If evoking currents in inhibitory cells the currents tend to to faster, while excitatory cells have slower currents.