The mistake here is thinking about this as an if statement, instead you want to use what people often call logical indexing. In general, you can take a vector and refer to a subset of it using another vector of false and true (or 0 and 1). And you can calculate those logical (true/false) vectors with comparison operators like <.
Here's a look at that in action with a similar dataset as what you describe:
X = randn(1000,1);
Y = randn(1000,1);
Var1 = rand(1000,1) * 1000; % so that some of the values are greater than 600 and some are not
figure(1)
scatter(X(Var1<600),Y(Var1<600),'r')
hold on
scatter(X(Var1>=600),Y(Var1>=600),'g')
Now let's do the same thing with a table:
figure(2)
t=table(X,Y,Var1);
scatter(t.X(t.Var1<600),t.Y(t.Var1<600),'r')
hold on
scatter(t.X(t.Var1>=600),t.Y(t.Var1>=600),'g')
Finally, I thought it might be worth pointing out, that scatter accepts a color input, and we could use that too so that there's just one call to scatter. This has the advantage that the points are plotted as they appear in the table, rather than plotting the green ones 'on top' of the red ones.
figure(3)
c=t.Var1<600;
scatter(t.X,t.Y,[],c)
colormap([1 0 0;0 1 0]); % scatter will use the colormap to choose colors for the values in c, so we need to tell it to use red and green
