Introduction:

I think the introduction could be expanded and there are some papers which
should be included.   Various IR emission-line diagnostics were proposed
prior to ISO; Voit (1992) and Spinoglio (1992) used photoionization models
to predict which line ratios would be useful in the IR.

More details on the contributions by the various authors (Voit, Spinoglio,
Genzel, Sturm, Laurent etc) in the IR diagnostics field would help the
introduction also.

Veilleux & Osterbrock (1987) should be added to the references for
emission-line diagnostic ratios.  Their paper was fundamental for
establishing the BPT optical emission-line diagnostics.

A paragraph describing the classifications obtained in optical compared
with the IR in previous work (Genzel, Taniguchi et al. 1999, etc) would be
useful.

--Thanks for the references.  I did reference work by Spinoglio and 
  Veilleux later in the paper, but it would be good to beef up the intro,
  too.

Section 3.3:
"We use the 24 micron data to normalize the line fluxes in some of our
diagnostic diagrams" - which diagrams and how reliable are the resultant
fluxes?

--The paragraph has been revised.  The 24um data are 10% uncertain.

Section 4.3:

The Kewley et al. diagnostics are not empirical and we did not use
thousands of galaxies to define our curves.  Our curves are purely
theoretical (we used stellar population synthesis and photoionization
models!) and our classification line gives the maximum position on the
optical diagnostic diagrams that can be obtained bu pure photoionization
models.  Objects lying above our line require an additional power source
such as an AGN or shocks.  Objects lying below our line may still contain
an AGN which may contribute up to ~30% of the emission-line flux ratios.
The Kauffmann line (which is empirical) aims to eliminate all galaxies
containing AGN, regardless of dominant power source.  Objects lying in
between the Kewley & Kauffmann lines are likely to be composite AGN-SF
objects but still dominated by star formation.

--The text has been revised accordingly.

It would be useful to include the [SII]/H-alpha and [OI]/H-alpha
diagnostic diagrams with our lines on them also because some galaxies lie
in the AGN region on one or both of these diagrams, but remain in the
starburst region on the [NII]/H-alpha diagram ([NII]/H-alpha not as
sensitive to a hard ionizing radiation field as [SII]/H-alpha or
[OI]/H-alpha because the latter are produced in the partially-ionized
zone).   Objects that lie in one region on one diagram and in another
region on another diagram may be composite objects also but are usually
classified as `ambiguous'.

--I have looked at these additional plots.  NGC 4552 is still out of place
  in the [SII]/Ha plot, but not in the [OI]/Ha version.  The other 
  galaxies mentioned in the text are either still 'troublesome' or close
  to your SF/SGN boundaries.  Since:
    -the focus of this paper is to explore 'new and existing' IR
    diagnostics, 
    -Kauffmann et al. don't provide similar curves for the [SII]/Ha and 
     [OI]/Ha plots, 
    -the [SII] and [OI] fluxes are weaker,
  I would prefer to just mention that we have looked at these other 
  optical diagnostics and do not significant differences.

A description of LINERs as a class would be useful, particularly since
LINERs are known to be a mixed bag of objects.  The lack of a clean
separation between LINERs and Seyferts is consistent with this.

--Done.

A more detailed description of the classifications is needed because the
classifications described in this section do not match the classifications
in Fig 3 onwards.  It is not clear what galaxies are classed as "Seyfert +
HII or LINER" or "LINER + HII or Starburst".

--I have simplified the figures to indicate only Seyfert, LINER, or
  starburst.

The last sentence "In the next section we turn to the mid-infrared where
we will perhaps more convincingle be able to determine whether the optical
classifications are appropriate" sounds weak and its not clear to me that
the paper has convincingly tested the optical classifications.  If this is
one of the main goals (or the main goal?) of the paper then more
discussion in the introduction on optical vs IR classifications is needed
and a more quantitative discussion whether the optical classifications are
consistent with the IR classifications is needed later on in Section 4.
If we are testing the optical diagnostics, it would be good to include
some more heavily reddened sources or atleast discuss them in more detail.
Because the SINGS sample only has modest dust attenuations, the referee
might question whether the SINGS sample is the most appropriate sample for
testing optical classifications. Maybe a different focus such as
'exploring new and existing IR diagnostics' would be better.

--I agree.  The SINGS sample is not optimally suited for testing optical
  vs infrared classifications.  So I have decreased the emphasis of this
  angle in the text.

Section 4.4

1st paragraph: The text refers to "normal nuclei" but its not really clear
what normal nuclei are.  Is this "pure starburst" nuclei?  low luminosity
starburst nuclei? it would be better to use similar wording to that used
in the figures.

--normal has been put in quotes and starbursting/star-forming has
  parenthetically been added to help clarify

2nd paragraph: The Maloney et al. quotes would probably read better if
paraphrased.

--Done

3rd paragraph: The discussion of the interstellar density interpretation
of the low [SiII]/[NeII] ratio in AGN is confusing.  The text says "if the
[SiII] comes mostly from X-ray dominated regions, then the low
[SiII]/[NeII] for AGn sources is due to strong [SiII] cooling.  When I
read "strong [SiII] cooling", I expect [SiII] emission to be high, and
therefore [SiII]/[NeII] would be higher, not lower.  Similarly the text
says "If most of the [SiII] emission comes from photo-dissociation
regions, then the low [SiII]/[NeII] for AGN sources is due to enhances
[SiII] emission in dense photo-dissociation regions." - but wouldn't an
enhanced [SiII] cause a higher [SiII]/[NeII] ratio not a lower one?

--These typos have been fixed.

5th paragraph: From the discussion in the 3rd paragraph, it sounds as if
the prominant cooling line from a low-excitation species (presumably
[SiII]) could come from either X-ray dominated regions or PDR regions, but
that we can't distinguish between the two.  If this is the case, then it
seems odd to then say that the new technique for distinguishing between
AGN sources and star-forming regions relies on a line associated with
X-ray dominated regions.  Couldn't it also come from PDR regions, as
described earlier?

--The wording has been improved to emphasize the continued possibility of
  the dense PDR scenario.

Figure 3: If either of these diagrams are going to be used to distinguish
starburst from AGN by other authors, it would be useful to provide a
classification line and formula in the paper.  It would also be useful to
quantify how many pure starbursts, pure AGN, and pure LINERs change
classes (zero?) using the new diagrams.  Are these fractions consistent
with those found by Genzel and others?  Lutz et al. found that optical and
IR classes agree well only if HII and LINERs are both classed as
starbursts in the IR.  Is this result consistent with the SINGS LINERs?

--I have added multiple classification lines in Figures 3 and 5, along
  with tables that quantify the percentage of types within each region.
  I don't find the Genzel reference that compares optical and IR classes;
  please send me the reference (and section).
  Finally, I would prefer to emphasize the new IR diagnostics in this
  paper.  But having said this, no "pure" systems changed classes from
  the optical diagram to those in Figures 3 and 5.

Discussion of Figure 4
Genzel's annual review on the ISO results gives a few references (Vermeij
& van der Hulst 1999, Cox et al. 1999) that show that [NeIII]/[NeII] ratio
depends strongly on metallicity.  It would be worth discussing the
metallicity effect that you are seeing in the context of these previous
results.

--I have added references to Genzel & Cesarsky (2000) and the relevant 
  peer-reviewed work that eventually came out of Vermeij's thesis 
  (Martin-Hernandez et al. 2002).

Section 4.4.3
It would be useful to provide the formula for the green line in Figure 5
so that others can easily use the new diagnostic.

--Done, plus I've added two more lines and a table on the statistics.

Pilyugin & Ferrini (1998) were not the first authors to show that the
central regions of galaxies are more abundant in heavy metals than their
outer regions.  Some key papers in this area are Pagel & Edmunds (1981),
McCall (1982), Vila-Costas & Edmunds (1992).  Henry & Worthey (1999) give
a good review of abundance gradients in spirals.

--The references have been added.  Thanks!

Section 4.4.4
"the nuclei with obvious AGN activity appear separately from the
starbursting nuclei and the HII regions".  How significant is the
difference given the errors?  Were errors used in determining the
least-squares fits to the data?
Which fits were "LINER = HII or Starburst" and "Seyfert = HII or LINER"
included in, if any?

--Seyferts+LINERS separate from star-forming sources along the diagonal. 
  I'm not saying that they have measurably different slopes.  The text has
  been cleared up to emphasize they seprate along the diagonal.

Summary
The discussion of the power of the Genzel diagram seems a bit out of
place, given that the previous sentence states that we don't have any
highly reddened sources and the IR diagnostics don't tell us anything
different from the optical.

--The sentence has been removed, and the summary revised.