diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c
new file mode 100644
index ddb68ab..0f93619
*** a/src/backend/commands/analyze.c
--- b/src/backend/commands/analyze.c
*************** compute_scalar_stats(VacAttrStatsP stats
*** 2464,2469 ****
--- 2464,2471 ----
  		 * emit duplicate histogram bin boundaries.  (We might end up with
  		 * duplicate histogram entries anyway, if the distribution is skewed;
  		 * but we prefer to treat such values as MCVs if at all possible.)
+ 		 * We also decrease ndistinct in the process such that going forward
+ 		 * it refers to the number of distinct values left for the histogram.
  		 */
  		if (track_cnt == ndistinct && toowide_cnt == 0 &&
  			stats->stadistinct > 0 &&
*************** compute_scalar_stats(VacAttrStatsP stats
*** 2471,2505 ****
  		{
  			/* Track list includes all values seen, and all will fit */
  			num_mcv = track_cnt;
  		}
  		else
  		{
! 			double		ndistinct = stats->stadistinct;
! 			double		avgcount,
! 						mincount,
! 						maxmincount;
  
- 			if (ndistinct < 0)
- 				ndistinct = -ndistinct * totalrows;
- 			/* estimate # of occurrences in sample of a typical value */
- 			avgcount = (double) samplerows / ndistinct;
- 			/* set minimum threshold count to store a value */
- 			mincount = avgcount * 1.25;
- 			if (mincount < 2)
- 				mincount = 2;
- 			/* don't let threshold exceed 1/K, however */
- 			maxmincount = (double) samplerows / (double) num_bins;
- 			if (mincount > maxmincount)
- 				mincount = maxmincount;
  			if (num_mcv > track_cnt)
  				num_mcv = track_cnt;
! 			for (i = 0; i < num_mcv; i++)
  			{
! 				if (track[i].count < mincount)
! 				{
! 					num_mcv = i;
  					break;
  				}
  			}
  		}
  
--- 2473,2533 ----
  		{
  			/* Track list includes all values seen, and all will fit */
  			num_mcv = track_cnt;
+ 
+ 			/* Nothing left for the histogram */
+ 			num_hist = 0;
+ 			ndistinct = 0;
  		}
  		else
  		{
! 			/*
! 			 * Starting number of values left for the histogram: samplerows
! 			 * sans nulls and too wide ones.
! 			 */
! 			int			sample_cnt = values_cnt;
  
  			if (num_mcv > track_cnt)
  				num_mcv = track_cnt;
! 			for (i = 0; /* i < num_mcv */; i++)
  			{
! 				/*
! 				 * We have to put this before the loop condition, otherwise
! 				 * we'll have to repeat this code before the loop and after
! 				 * decreasing ndistinct.
! 				 */
! 				num_hist = ndistinct;
! 				if (num_hist > num_bins)
! 					num_hist = num_bins + 1;
! 
! 				/* Another way to say "while (i < num_mcv)" */
! 				if (i >= num_mcv)
  					break;
+ 
+ 				if (num_hist >= 2)
+ 				{
+ 					double		avgcount,
+ 								mincount,
+ 								maxmincount;
+ 
+ 					/* estimate # of occurrences in sample of a typical value */
+ 					avgcount = (double) sample_cnt / (double) ndistinct;
+ 
+ 					/* set minimum threshold count to store a value */
+ 					mincount = 1.25 * avgcount;
+ 
+ 					/* don't let threshold exceed 1/K, however */
+ 					maxmincount = (sample_cnt - 1) / (double) (num_hist - 1);
+ 					if (mincount > maxmincount)
+ 						mincount = maxmincount;
+ 					if (track[i].count < mincount)
+ 					{
+ 						num_mcv = i;
+ 						break;
+ 					}
  				}
+ 				/* Narrow our view of samples left for the histogram */
+ 				sample_cnt -= track[i].count;
+ 				ndistinct--;
  			}
  		}
  
*************** compute_scalar_stats(VacAttrStatsP stats
*** 2542,2550 ****
  		 * values not accounted for in the MCV list.  (This ensures the
  		 * histogram won't collapse to empty or a singleton.)
  		 */
- 		num_hist = ndistinct - num_mcv;
- 		if (num_hist > num_bins)
- 			num_hist = num_bins + 1;
  		if (num_hist >= 2)
  		{
  			MemoryContext old_context;
--- 2570,2575 ----