The overwhelming majority of non-severe and severe storms throughout the contiguous U.S. generate primarily (> 75%) negative ground flashes (so-called negative storms).  However, a certain subset of severe storms produces an anomalously high (> 25%) percentage of positive ground flashes (so-called positive storms).  The frequency of these “anomalous” positive storms varies regionally and seasonally.  In some regions (e.g., central and northern plains) and months, these positive storms are common, representing 30% or more of all severe storms.    Enhanced positive CG production in some severe storms has led to the idea that CG lightning polarity may be useful in the short-term prediction of severe weather.  Before implementing such a nowcasting tool, we must first better understand the relationship between the mesoscale environment, storm intensity and severity, and CG lightning polarity. Several past studies have noted that severe storms passing through similar mesoscale regions on a given day exhibit similar CG lightning behavior.  This repeated observation led to the hypothesis that the local mesoscale environment indirectly influences CG lightning polarity by directly controlling storm structure, dynamics, and microphysics, which in turn control storm electrification.  According to the hypothesis, intense updrafts and associated high liquid water contents in positive storms lead to positive charging of graupel and hail via the non-inductive charging mechanism, an enhanced lower positive charge, and increased frequency of positive CG lightning.  A handful of studies have explored the relationship between the local mesoscale environment and the CG lightning behavior of severe storms.  Since it is difficult to obtain representative soundings, further study is warranted. We have utilized abundant environmental soundings taken during the International H₂O project (IHOP, May-June 2002) to document the relationship between mesoscale environment and dominant CG lightning polarity.  We identified one non-severe negative (23 May), four severe negative (24 May; 4, 12, 15 June), and four severe positive (23, 24 May; 15, 19 June) storm systems on six different days during IHOP.   From hundreds of IHOP soundings, we carefully selected roughly fifty inflow proximity soundings that best represented the mesoscale environment of the nine storm systems.  Consistent with past results, deep layer (0-6 km) shear had little control on CG lightning polarity.  At odds with past studies, we found little difference in mean CAPE between positive and negative severe storms.  On the other hand, a modified mean CAPE estimated in the electrically important mixed phase zone from the -10°C to the -40°C level (or mixed phase CAPE) was significantly higher in positive storms (1210 J kg^-1) than in negative storms (957 J kg^-1) (statistically significant at the 5% level).  Interestingly, the average LCL for positive storms (2079 m) was 1.9 times higher than for negative storms (1121 m) (significant to the 0.1% level).  Furthermore, the environmental freezing level was lower in positive storms (3777 m) than in negative storms (4070 m) (significant to the 1% level).  Combining these last two results, the mean warm cloud depth, which is defined as the distance from the cloud base or LCL to the freezing level, was dramatically bigger in negative storms (2949 m) than in positive storms (1699 m) (significant to the 0.01% level).  In the figure below, we show the relationship between the median mixed phase CAPE, warm cloud depth, low-level (0-3 km) shear, and dominant CG lightning polarity.  The size of the circle and the interior value below indicate the strength of the 0-3 km shear (bigger circle=larger low-level shear).  Consistent with past work, positive storms (14.7 m s^-1) had noticeably higher mean low-level shear than negative storms (10.7 m s^-1) (significant to the 1% level).  However, differences in warm cloud depth, more than any other environmental factor, most clearly separated positive storms from negative ones. The results above are consistent with the hypothesis that intense updrafts and high liquid water contents in the mixed phase zone, where cloud electrification is likely occurring via the non-inductive charging mechanism, are correlated with enhanced positive CG lightning production.  According to parcel theory, higher mixed phase CAPE directly leads to stronger updrafts and higher liquid water contents in positive storms.  Larger low-level shear in positive storms aids in the development of intense low-to-mid level updrafts and enhanced liquid water contents through well known dynamical forcing.  Higher LCL or cloud base height, which is associated with increased parcel size and decreased entrainment of dry air, in positive storms may lead to more efficient conversion of CAPE into kinetic energy and hence enhanced updraft strength and liquid water content.  Furthermore, significantly reduced warm cloud depth in positive storms may decrease the amount of liquid water that is lost through the collision-coalescence and rainout process in a rising air parcel below the mixed phase zone, effectively increasing the amount of supercooled cloud water that is available in the mixed phase zone for cloud electrification.  Further analysis and interpretation of IHOP sounding data along with implications of these results for cloud electrification, lightning, and the nowcasting of severe storms using CG polarity information will be presented.